Jens Behley

Tim-Felix Faasch, Jochen Kall, Lucas Nunes et al.

High-fidelity semantic 3D scene representations are crucial for numerous applications, including robotics, autonomous driving, and simulation. Beyond this, the ability to edit such representations enables developers to adapt these applications more easily to specific target scenarios. Current approaches provide limited support for controllable editing. We introduce TASE, a method that projects pretrained 2D semantic features into a truncation-aware embedding space to enable flexible 3D scene editing. Our method explicitly optimizes a feature space in which progressively reducing feature channels yields increasingly abstract semantic representations, while retaining more channels preserves fine-grained detail. Additionally, we improve multi-view consistency of the features using a scale- and translation-equivariance loss. The resulting truncation-aware embedding space enables text-driven edits to 3D scenes, providing explicit control over how strongly edits adhere to the original scene content and allowing more substantial modifications than prior methods. Moreover, we propose a finetuning stage for the editing diffusion model to mitigate artifacts caused by geometric changes. Experimental results demonstrate competitive performance in 3D scene editing, substantially outperforming prior methods on edits involving large geometric modifications.

Yue Pan, Federico Magistri, Thomas Läbe et al.

Monitoring plants and fruits at high resolution play a key role in the future of agriculture. Accurate 3D information can pave the way to a diverse number of robotic applications in agriculture ranging from autonomous harvesting to precise yield estimation. Obtaining such 3D information is non-trivial as agricultural environments are often repetitive and cluttered, and one has to account for the partial observability of fruit and plants. In this paper, we address the problem of jointly estimating complete 3D shapes of fruit and their pose in a 3D multi-resolution map built by a mobile robot. To this end, we propose an online multi-resolution panoptic mapping system where regions of interest are represented with a higher resolution. We exploit data to learn a general fruit shape representation that we use at inference time together with an occlusion-aware differentiable rendering pipeline to complete partial fruit observations and estimate the 7 DoF pose of each fruit in the map. The experiments presented in this paper evaluated both in the controlled environment and in a commercial greenhouse, show that our novel algorithm yields higher completion and pose estimation accuracy than existing methods, with an improvement of 41% in completion accuracy and 52% in pose estimation accuracy while keeping a low inference time of 0.6s in average. Codes are available at: https://github.com/PRBonn/HortiMapping.

Haofei Kuang, Xieyuanli Chen, Tiziano Guadagnino et al.

Determining the state of a mobile robot is an essential building block of robot navigation systems. In this paper, we address the problem of estimating the robots pose in an indoor environment using 2D LiDAR data and investigate how modern environment models can improve gold standard Monte-Carlo localization (MCL) systems. We propose a neural occupancy field to implicitly represent the scene using a neural network. With the pretrained network, we can synthesize 2D LiDAR scans for an arbitrary robot pose through volume rendering. Based on the implicit representation, we can obtain the similarity between a synthesized and actual scan as an observation model and integrate it into an MCL system to perform accurate localization. We evaluate our approach on self-recorded datasets and three publicly available ones. We show that we can accurately and efficiently localize a robot using our approach surpassing the localization performance of state-of-the-art methods. The experiments suggest that the presented implicit representation is able to predict more accurate 2D LiDAR scans leading to an improved observation model for our particle filter-based localization. The code of our approach will be available at: https://github.com/PRBonn/ir-mcl.

Nicky Zimmerman, Matteo Sodano, Elias Marks et al.

Object-based maps are relevant for scene understanding since they integrate geometric and semantic information of the environment, allowing autonomous robots to robustly localize and interact with on objects. In this paper, we address the task of constructing a metric-semantic map for the purpose of long-term object-based localization. We exploit 3D object detections from monocular RGB frames for both, the object-based map construction, and for globally localizing in the constructed map. To tailor the approach to a target environment, we propose an efficient way of generating 3D annotations to finetune the 3D object detection model. We evaluate our map construction in an office building, and test our long-term localization approach on challenging sequences recorded in the same environment over nine months. The experiments suggest that our approach is suitable for constructing metric-semantic maps, and that our localization approach is robust to long-term changes. Both, the mapping algorithm and the localization pipeline can run online on an onboard computer. We release an open-source C++/ROS implementation of our approach.

Yue Pan, Tao Sun, Liyuan Zhu et al.

Point cloud registration aligns multiple unposed point clouds into a common frame, and is a core step for 3D reconstruction and robot localization. In this work, we cast registration as conditional generation: a learned continuous, point-wise velocity field transports noisy points to a registered scene, from which the pose of each view is recovered. Unlike previous methods that conduct correspondence matching to estimate the transformation between a pair of point clouds and then optimize the pairwise transformations to realize multi-view registration, our model directly generates the registered point cloud. With a lightweight local feature extractor and test-time rigidity enforcement, our approach achieves state-of-the-art results on pairwise and multi-view registration benchmarks, particularly with low overlap, and generalizes across scales and sensor modalities. It further supports downstream tasks including relocalization, multi-robot SLAM, and multi-session map merging. Source code available at: https://github.com/PRBonn/RAP.

Benedikt Mersch, Xieyuanli Chen, Ignacio Vizzo et al.

A key challenge for autonomous vehicles is to navigate in unseen dynamic environments. Separating moving objects from static ones is essential for navigation, pose estimation, and understanding how other traffic participants are likely to move in the near future. In this work, we tackle the problem of distinguishing 3D LiDAR points that belong to currently moving objects, like walking pedestrians or driving cars, from points that are obtained from non-moving objects, like walls but also parked cars. Our approach takes a sequence of observed LiDAR scans and turns them into a voxelized sparse 4D point cloud. We apply computationally efficient sparse 4D convolutions to jointly extract spatial and temporal features and predict moving object confidence scores for all points in the sequence. We develop a receding horizon strategy that allows us to predict moving objects online and to refine predictions on the go based on new observations. We use a binary Bayes filter to recursively integrate new predictions of a scan resulting in more robust estimation. We evaluate our approach on the SemanticKITTI moving object segmentation challenge and show more accurate predictions than existing methods. Since our approach only operates on the geometric information of point clouds over time, it generalizes well to new, unseen environments, which we evaluate on the Apollo dataset.

Xingguang Zhong, Yue Pan, Jens Behley et al.

Accurate mapping of large-scale environments is an essential building block of most outdoor autonomous systems. Challenges of traditional mapping methods include the balance between memory consumption and mapping accuracy. This paper addresses the problem of achieving large-scale 3D reconstruction using implicit representations built from 3D LiDAR measurements. We learn and store implicit features through an octree-based, hierarchical structure, which is sparse and extensible. The implicit features can be turned into signed distance values through a shallow neural network. We leverage binary cross entropy loss to optimize the local features with the 3D measurements as supervision. Based on our implicit representation, we design an incremental mapping system with regularization to tackle the issue of forgetting in continual learning. Our experiments show that our 3D reconstructions are more accurate, complete, and memory-efficient than current state-of-the-art 3D mapping methods.

Marco Camurri, Enrico Tomelleri, Matías Mattamala et al. · oxford

Covering one third of Earth's land surface, forests are vital to global biodiversity, climate regulation, and human well-being. In Europe, forests and woodlands reach approximately 40% of land area, and the forestry sector is central to achieving the EU's climate neutrality and biodiversity goals; these emphasize sustainable forest management, increased use of long-lived wood products, and resilient forest ecosystems. To meet these goals and properly address their inherent challenges, current practices require further innovation. This chapter introduces DigiForest, a novel, large-scale precision forestry approach leveraging digital technologies and autonomous robotics. DigiForest is structured around four main components: (1) autonomous, heterogeneous mobile robots (aerial, legged, and marsupial) for tree-level data collection; (2) automated extraction of tree traits to build forest inventories; (3) a Decision Support System (DSS) for forecasting forest growth and supporting decision-making; and (4) low-impact selective logging using purpose-built autonomous harvesters. These technologies have been extensively validated in real-world conditions in several locations, including forests in Finland, the UK, and Switzerland.

Jan Weyler, Federico Magistri, Elias Marks et al.

The production of food, feed, fiber, and fuel is a key task of agriculture, which has to cope with many challenges in the upcoming decades, e.g., a higher demand, climate change, lack of workers, and the availability of arable land. Vision systems can support making better and more sustainable field management decisions, but also support the breeding of new crop varieties by allowing temporally dense and reproducible measurements. Recently, agricultural robotics got an increasing interest in the vision and robotics communities since it is a promising avenue for coping with the aforementioned lack of workers and enabling more sustainable production. While large datasets and benchmarks in other domains are readily available and enable significant progress, agricultural datasets and benchmarks are comparably rare. We present an annotated dataset and benchmarks for the semantic interpretation of real agricultural fields. Our dataset recorded with a UAV provides high-quality, pixel-wise annotations of crops and weeds, but also crop leaf instances at the same time. Furthermore, we provide benchmarks for various tasks on a hidden test set comprised of different fields: known fields covered by the training data and a completely unseen field. Our dataset, benchmarks, and code are available at \url{https://www.phenobench.org}.

Gianmarco Roggiolani, Matteo Sodano, Tiziano Guadagnino et al.

Plant phenotyping is a central task in agriculture, as it describes plants' growth stage, development, and other relevant quantities. Robots can help automate this process by accurately estimating plant traits such as the number of leaves, leaf area, and the plant size. In this paper, we address the problem of joint semantic, plant instance, and leaf instance segmentation of crop fields from RGB data. We propose a single convolutional neural network that addresses the three tasks simultaneously, exploiting their underlying hierarchical structure. We introduce task-specific skip connections, which our experimental evaluation proves to be more beneficial than the usual schemes. We also propose a novel automatic post-processing, which explicitly addresses the problem of spatially close instances, common in the agricultural domain because of overlapping leaves. Our architecture simultaneously tackles these problems jointly in the agricultural context. Previous works either focus on plant or leaf segmentation, or do not optimise for semantic segmentation. Results show that our system has superior performance compared to state-of-the-art approaches, while having a reduced number of parameters and is operating at camera frame rate.

Matthias Zeller, Jens Behley, Michael Heidingsfeld et al.

Scene understanding is crucial for autonomous robots in dynamic environments for making future state predictions, avoiding collisions, and path planning. Camera and LiDAR perception made tremendous progress in recent years, but face limitations under adverse weather conditions. To leverage the full potential of multi-modal sensor suites, radar sensors are essential for safety critical tasks and are already installed in most new vehicles today. In this paper, we address the problem of semantic segmentation of moving objects in radar point clouds to enhance the perception of the environment with another sensor modality. Instead of aggregating multiple scans to densify the point clouds, we propose a novel approach based on the self-attention mechanism to accurately perform sparse, single-scan segmentation. Our approach, called Gaussian Radar Transformer, includes the newly introduced Gaussian transformer layer, which replaces the softmax normalization by a Gaussian function to decouple the contribution of individual points. To tackle the challenge of the transformer to capture long-range dependencies, we propose our attentive up- and downsampling modules to enlarge the receptive field and capture strong spatial relations. We compare our approach to other state-of-the-art methods on the RadarScenes data set and show superior segmentation quality in diverse environments, even without exploiting temporal information.

Matthias Zeller, Vardeep S. Sandhu, Benedikt Mersch et al.

The perception of moving objects is crucial for autonomous robots performing collision avoidance in dynamic environments. LiDARs and cameras tremendously enhance scene interpretation but do not provide direct motion information and face limitations under adverse weather. Radar sensors overcome these limitations and provide Doppler velocities, delivering direct information on dynamic objects. In this paper, we address the problem of moving instance segmentation in radar point clouds to enhance scene interpretation for safety-critical tasks. Our Radar Instance Transformer enriches the current radar scan with temporal information without passing aggregated scans through a neural network. We propose a full-resolution backbone to prevent information loss in sparse point cloud processing. Our instance transformer head incorporates essential information to enhance segmentation but also enables reliable, class-agnostic instance assignments. In sum, our approach shows superior performance on the new moving instance segmentation benchmarks, including diverse environments, and provides model-agnostic modules to enhance scene interpretation. The benchmark is based on the RadarScenes dataset and will be made available upon acceptance.

Matteo Sodano, Federico Magistri, Tiziano Guadagnino et al.

Perception is crucial for robots that act in real-world environments, as autonomous systems need to see and understand the world around them to act properly. Panoptic segmentation provides an interpretation of the scene by computing a pixelwise semantic label together with instance IDs. In this paper, we address panoptic segmentation using RGB-D data of indoor scenes. We propose a novel encoder-decoder neural network that processes RGB and depth separately through two encoders. The features of the individual encoders are progressively merged at different resolutions, such that the RGB features are enhanced using complementary depth information. We propose a novel merging approach called ResidualExcite, which reweighs each entry of the feature map according to its importance. With our double-encoder architecture, we are robust to missing cues. In particular, the same model can train and infer on RGB-D, RGB-only, and depth-only input data, without the need to train specialized models. We evaluate our method on publicly available datasets and show that our approach achieves superior results compared to other common approaches for panoptic segmentation.

Hyungtae Lim, Seoyeon Jang, Benedikt Mersch et al.

Moving object segmentation (MOS) using a 3D light detection and ranging (LiDAR) sensor is crucial for scene understanding and identification of moving objects. Despite the availability of various types of 3D LiDAR sensors in the market, MOS research still predominantly focuses on 3D point clouds from mechanically spinning omnidirectional LiDAR sensors. Thus, we are, for example, lacking a dataset with MOS labels for point clouds from solid-state LiDAR sensors which have irregular scanning patterns. In this paper, we present a labeled dataset, called \textit{HeLiMOS}, that enables to test MOS approaches on four heterogeneous LiDAR sensors, including two solid-state LiDAR sensors. Furthermore, we introduce a novel automatic labeling method to substantially reduce the labeling effort required from human annotators. To this end, our framework exploits an instance-aware static map building approach and tracking-based false label filtering. Finally, we provide experimental results regarding the performance of commonly used state-of-the-art MOS approaches on HeLiMOS that suggest a new direction for a sensor-agnostic MOS, which generally works regardless of the type of LiDAR sensors used to capture 3D point clouds. Our dataset is available at https://sites.google.com/view/helimos.

Gianmarco Roggiolani, Federico Magistri, Tiziano Guadagnino et al.

Agricultural robots have the prospect to enable more efficient and sustainable agricultural production of food, feed, and fiber. Perception of crops and weeds is a central component of agricultural robots that aim to monitor fields and assess the plants as well as their growth stage in an automatic manner. Semantic perception mostly relies on deep learning using supervised approaches, which require time and qualified workers to label fairly large amounts of data. In this paper, we look into the problem of reducing the amount of labels without compromising the final segmentation performance. For robots operating in the field, pre-training networks in a supervised way is already a popular method to reduce the number of required labeled images. We investigate the possibility of pre-training in a self-supervised fashion using data from the target domain. To better exploit this data, we propose a set of domain-specific augmentation strategies. We evaluate our pre-training on semantic segmentation and leaf instance segmentation, two important tasks in our domain. The experimental results suggest that pre-training with domain-specific data paired with our data augmentation strategy leads to superior performance compared to commonly used pre-trainings. Furthermore, the pre-trained networks obtain similar performance to the fully supervised with less labeled data.

Federico Magistri, Thomas Läbe, Elias Marks et al.

As the world population is expected to reach 10 billion by 2050, our agricultural production system needs to double its productivity despite a decline of human workforce in the agricultural sector. Autonomous robotic systems are one promising pathway to increase productivity by taking over labor-intensive manual tasks like fruit picking. To be effective, such systems need to monitor and interact with plants and fruits precisely, which is challenging due to the cluttered nature of agricultural environments causing, for example, strong occlusions. Thus, being able to estimate the complete 3D shapes of objects in presence of occlusions is crucial for automating operations such as fruit harvesting. In this paper, we propose the first publicly available 3D shape completion dataset for agricultural vision systems. We provide an RGB-D dataset for estimating the 3D shape of fruits. Specifically, our dataset contains RGB-D frames of single sweet peppers in lab conditions but also in a commercial greenhouse. For each fruit, we additionally collected high-precision point clouds that we use as ground truth. For acquiring the ground truth shape, we developed a measuring process that allows us to record data of real sweet pepper plants, both in the lab and in the greenhouse with high precision, and determine the shape of the sensed fruits. We release our dataset, consisting of almost 7,000 RGB-D frames belonging to more than 100 different fruits. We provide segmented RGB-D frames, with camera intrinsics to easily obtain colored point clouds, together with the corresponding high-precision, occlusion-free point clouds obtained with a high-precision laser scanner. We additionally enable evaluation of shape completion approaches on a hidden test set through a public challenge on a benchmark server.

Xingguang Zhong, Liren Jin, Marija Popović et al.

Reliable incremental estimation of camera poses and 3D reconstruction is key to enable various applications including robotics, interactive visualization, and augmented reality. However, this task is particularly challenging in dynamic natural environments, where scene dynamics can severely deteriorate camera pose estimation accuracy. In this work, we propose a novel monocular visual SLAM system that can robustly estimate camera poses in dynamic scenes. To this end, we leverage the complementary strengths of geometric patch-based online bundle adjustment and recent feed-forward reconstruction models. Specifically, we propose a feed-forward reconstruction model to precisely filter out dynamic regions, while also utilizing its depth prediction to enhance the robustness of the patch-based visual SLAM. By aligning depth prediction with estimated patches from bundle adjustment, we robustly handle the inherent scale ambiguities of the batch-wise application of the feed-forward reconstruction model.

Yue Pan, Xingguang Zhong, Louis Wiesmann et al.

Accurate and robust localization and mapping are essential components for most autonomous robots. In this paper, we propose a SLAM system for building globally consistent maps, called PIN-SLAM, that is based on an elastic and compact point-based implicit neural map representation. Taking range measurements as input, our approach alternates between incremental learning of the local implicit signed distance field and the pose estimation given the current local map using a correspondence-free, point-to-implicit model registration. Our implicit map is based on sparse optimizable neural points, which are inherently elastic and deformable with the global pose adjustment when closing a loop. Loops are also detected using the neural point features. Extensive experiments validate that PIN-SLAM is robust to various environments and versatile to different range sensors such as LiDAR and RGB-D cameras. PIN-SLAM achieves pose estimation accuracy better or on par with the state-of-the-art LiDAR odometry or SLAM systems and outperforms the recent neural implicit SLAM approaches while maintaining a more consistent, and highly compact implicit map that can be reconstructed as accurate and complete meshes. Finally, thanks to the voxel hashing for efficient neural points indexing and the fast implicit map-based registration without closest point association, PIN-SLAM can run at the sensor frame rate on a moderate GPU. Codes will be available at: https://github.com/PRBonn/PIN_SLAM.

Yue Pan, Xingguang Zhong, Liren Jin et al.

Robots benefit from high-fidelity reconstructions of their environment, which should be geometrically accurate and photorealistic to support downstream tasks. While this can be achieved by building distance fields from range sensors and radiance fields from cameras, realising scalable incremental mapping of both fields consistently and at the same time with high quality is challenging. In this paper, we propose a novel map representation that unifies a continuous signed distance field and a Gaussian splatting radiance field within an elastic and compact point-based implicit neural map. By enforcing geometric consistency between these fields, we achieve mutual improvements by exploiting both modalities. We present a novel LiDAR-visual SLAM system called PINGS using the proposed map representation and evaluate it on several challenging large-scale datasets. Experimental results demonstrate that PINGS can incrementally build globally consistent distance and radiance fields encoded with a compact set of neural points. Compared to state-of-the-art methods, PINGS achieves superior photometric and geometric rendering at novel views by constraining the radiance field with the distance field. Furthermore, by utilizing dense photometric cues and multi-view consistency from the radiance field, PINGS produces more accurate distance fields, leading to improved odometry estimation and mesh reconstruction. We also provide an open-source implementation of PING at: https://github.com/PRBonn/PINGS.

Matteo Sodano, Federico Magistri, Jens Behley et al.

Perception is a key building block of autonomously acting vision systems such as autonomous vehicles. It is crucial that these systems are able to understand their surroundings in order to operate safely and robustly. Additionally, autonomous systems deployed in unconstrained real-world scenarios must be able of dealing with novel situations and object that have never been seen before. In this article, we tackle the problem of open-world panoptic segmentation, i.e., the task of discovering new semantic categories and new object instances at test time, while enforcing consistency among the categories that we incrementally discover. We propose Con2MAV, an approach for open-world panoptic segmentation that extends our previous work, ContMAV, which was developed for open-world semantic segmentation. Through extensive experiments across multiple datasets, we show that our model achieves state-of-the-art results on open-world segmentation tasks, while still performing competitively on the known categories. We will open-source our implementation upon acceptance. Additionally, we propose PANIC (Panoptic ANomalies In Context), a benchmark for evaluating open-world panoptic segmentation in autonomous driving scenarios. This dataset, recorded with a multi-modal sensor suite mounted on a car, provides high-quality, pixel-wise annotations of anomalous objects at both semantic and instance level. Our dataset contains 800 images, with more than 50 unknown classes, i.e., classes that do not appear in the training set, and 4000 object instances, making it an extremely challenging dataset for open-world segmentation tasks in the autonomous driving scenario. We provide competitions for multiple open-world tasks on a hidden test set. Our dataset and competitions are available at https://www.ipb.uni-bonn.de/data/panic.

Lucas Nunes, Rodrigo Marcuzzi, Jens Behley et al.

Semantic scene understanding is crucial for robotics and computer vision applications. In autonomous driving, 3D semantic segmentation plays an important role for enabling safe navigation. Despite significant advances in the field, the complexity of collecting and annotating 3D data is a bottleneck in this developments. To overcome that data annotation limitation, synthetic simulated data has been used to generate annotated data on demand. There is still however a domain gap between real and simulated data. More recently, diffusion models have been in the spotlight, enabling close-to-real data synthesis. Those generative models have been recently applied to the 3D data domain for generating scene-scale data with semantic annotations. Still, those methods either rely on image projection or decoupled models trained with different resolutions in a coarse-to-fine manner. Such intermediary representations impact the generated data quality due to errors added in those transformations. In this work, we propose a novel approach able to generate 3D semantic scene-scale data without relying on any projection or decoupled trained multi-resolution models, achieving more realistic semantic scene data generation compared to previous state-of-the-art methods. Besides improving 3D semantic scene-scale data synthesis, we thoroughly evaluate the use of the synthetic scene samples as labeled data to train a semantic segmentation network. In our experiments, we show that using the synthetic annotated data generated by our method as training data together with the real semantic segmentation labels, leads to an improvement in the semantic segmentation model performance. Our results show the potential of generated scene-scale point clouds to generate more training data to extend existing datasets, reducing the data annotation effort. Our code is available at https://github.com/PRBonn/3DiSS.

Matteo Sodano, Federico Magistri, Elias Marks et al.

Crop yield estimation is a relevant problem in agriculture, because an accurate yield estimate can support farmers' decisions on harvesting or precision intervention. Robots can help to automate this process. To do so, they need to be able to perceive the surrounding environment to identify target objects such as trees and plants. In this paper, we introduce a novel approach to address the problem of hierarchical panoptic segmentation of apple orchards on 3D data from different sensors. Our approach is able to simultaneously provide semantic segmentation, instance segmentation of trunks and fruits, and instance segmentation of trees (a trunk with its fruits). This allows us to identify relevant information such as individual plants, fruits, and trunks, and capture the relationship among them, such as precisely estimate the number of fruits associated to each tree in an orchard. To efficiently evaluate our approach for hierarchical panoptic segmentation, we provide a dataset designed specifically for this task. Our dataset is recorded in Bonn, Germany, in a real apple orchard with a variety of sensors, spanning from a terrestrial laser scanner to a RGB-D camera mounted on different robots platforms. The experiments show that our approach surpasses state-of-the-art approaches in 3D panoptic segmentation in the agricultural domain, while also providing full hierarchical panoptic segmentation. Our dataset is publicly available at https://www.ipb.uni-bonn.de/data/hops/. The open-source implementation of our approach is available at https://github.com/PRBonn/hapt3D.

Xingguang Zhong, Yue Pan, Cyrill Stachniss et al.

Building accurate maps is a key building block to enable reliable localization, planning, and navigation of autonomous vehicles. We propose a novel approach for building accurate maps of dynamic environments utilizing a sequence of LiDAR scans. To this end, we propose encoding the 4D scene into a novel spatio-temporal implicit neural map representation by fitting a time-dependent truncated signed distance function to each point. Using our representation, we extract the static map by filtering the dynamic parts. Our neural representation is based on sparse feature grids, a globally shared decoder, and time-dependent basis functions, which we jointly optimize in an unsupervised fashion. To learn this representation from a sequence of LiDAR scans, we design a simple yet efficient loss function to supervise the map optimization in a piecewise way. We evaluate our approach on various scenes containing moving objects in terms of the reconstruction quality of static maps and the segmentation of dynamic point clouds. The experimental results demonstrate that our method is capable of removing the dynamic part of the input point clouds while reconstructing accurate and complete 3D maps, outperforming several state-of-the-art methods. Codes are available at: https://github.com/PRBonn/4dNDF

Emanuele Palazzolo, Jens Behley, Philipp Lottes et al.

Mapping and localization are essential capabilities of robotic systems. Although the majority of mapping systems focus on static environments, the deployment in real-world situations requires them to handle dynamic objects. In this paper, we propose an approach for an RGB-D sensor that is able to consistently map scenes containing multiple dynamic elements. For localization and mapping, we employ an efficient direct tracking on the truncated signed distance function (TSDF) and leverage color information encoded in the TSDF to estimate the pose of the sensor. The TSDF is efficiently represented using voxel hashing, with most computations parallelized on a GPU. For detecting dynamics, we exploit the residuals obtained after an initial registration, together with the explicit modeling of free space in the model. We evaluate our approach on existing datasets, and provide a new dataset showing highly dynamic scenes. These experiments show that our approach often surpass other state-of-the-art dense SLAM methods. We make available our dataset with the ground truth for both the trajectory of the RGB-D sensor obtained by a motion capture system and the model of the static environment using a high-precision terrestrial laser scanner. Finally, we release our approach as open source code.

Lucas Nunes, Rodrigo Marcuzzi, Benedikt Mersch et al.

Computer vision techniques play a central role in the perception stack of autonomous vehicles. Such methods are employed to perceive the vehicle surroundings given sensor data. 3D LiDAR sensors are commonly used to collect sparse 3D point clouds from the scene. However, compared to human perception, such systems struggle to deduce the unseen parts of the scene given those sparse point clouds. In this matter, the scene completion task aims at predicting the gaps in the LiDAR measurements to achieve a more complete scene representation. Given the promising results of recent diffusion models as generative models for images, we propose extending them to achieve scene completion from a single 3D LiDAR scan. Previous works used diffusion models over range images extracted from LiDAR data, directly applying image-based diffusion methods. Distinctly, we propose to directly operate on the points, reformulating the noising and denoising diffusion process such that it can efficiently work at scene scale. Together with our approach, we propose a regularization loss to stabilize the noise predicted during the denoising process. Our experimental evaluation shows that our method can complete the scene given a single LiDAR scan as input, producing a scene with more details compared to state-of-the-art scene completion methods. We believe that our proposed diffusion process formulation can support further research in diffusion models applied to scene-scale point cloud data.

Matteo Sodano, Federico Magistri, Lucas Nunes et al.

Interpreting camera data is key for autonomously acting systems, such as autonomous vehicles. Vision systems that operate in real-world environments must be able to understand their surroundings and need the ability to deal with novel situations. This paper tackles open-world semantic segmentation, i.e., the variant of interpreting image data in which objects occur that have not been seen during training. We propose a novel approach that performs accurate closed-world semantic segmentation and, at the same time, can identify new categories without requiring any additional training data. Our approach additionally provides a similarity measure for every newly discovered class in an image to a known category, which can be useful information in downstream tasks such as planning or mapping. Through extensive experiments, we show that our model achieves state-of-the-art results on classes known from training data as well as for anomaly segmentation and can distinguish between different unknown classes.

Liren Jin, Xingguang Zhong, Yue Pan et al.

Robotics applications often rely on scene reconstructions to enable downstream tasks. In this work, we tackle the challenge of actively building an accurate map of an unknown scene using an RGB-D camera on a mobile platform. We propose a hybrid map representation that combines a Gaussian splatting map with a coarse voxel map, leveraging the strengths of both representations: the high-fidelity scene reconstruction capabilities of Gaussian splatting and the spatial modelling strengths of the voxel map. At the core of our framework is an effective confidence modelling technique for the Gaussian splatting map to identify under-reconstructed areas, while utilising spatial information from the voxel map to target unexplored areas and assist in collision-free path planning. By actively collecting scene information in under-reconstructed and unexplored areas for map updates, our approach achieves superior Gaussian splatting reconstruction results compared to state-of-the-art approaches. Additionally, we demonstrate the real-world applicability of our framework using an unmanned aerial vehicle.

Matthias Zeller, Vardeep S. Sandhu, Benedikt Mersch et al.

The awareness about moving objects in the surroundings of a self-driving vehicle is essential for safe and reliable autonomous navigation. The interpretation of LiDAR and camera data achieves exceptional results but typically requires to accumulate and process temporal sequences of data in order to extract motion information. In contrast, radar sensors, which are already installed in most recent vehicles, can overcome this limitation as they directly provide the Doppler velocity of the detections and, hence incorporate instantaneous motion information within a single measurement. % In this paper, we tackle the problem of moving object segmentation in noisy radar point clouds. We also consider differentiating parked from moving cars, to enhance scene understanding. Instead of exploiting temporal dependencies to identify moving objects, we develop a novel transformer-based approach to perform single-scan moving object segmentation in sparse radar scans accurately. The key to our Radar Velocity Transformer is to incorporate the valuable velocity information throughout each module of the network, thereby enabling the precise segmentation of moving and non-moving objects. Additionally, we propose a transformer-based upsampling, which enhances the performance by adaptively combining information and overcoming the limitation of interpolation of sparse point clouds. Finally, we create a new radar moving object segmentation benchmark based on the RadarScenes dataset and compare our approach to other state-of-the-art methods. Our network runs faster than the frame rate of the sensor and shows superior segmentation results using only single-scan radar data.

Matthias Zeller, Daniel Casado Herraez, Jens Behley et al.

Robots and autonomous vehicles should be aware of what happens in their surroundings. The segmentation and tracking of moving objects are essential for reliable path planning, including collision avoidance. We investigate this estimation task for vehicles using radar sensing. We address moving instance tracking in sparse radar point clouds to enhance scene interpretation. We propose a learning-based radar tracker incorporating temporal offset predictions to enable direct center-based association and enhance segmentation performance by including additional motion cues. We implement attention-based tracking for sparse radar scans to include appearance features and enhance performance. The final association combines geometric and appearance features to overcome the limitations of center-based tracking to associate instances reliably. Our approach shows an improved performance on the moving instance tracking benchmark of the RadarScenes dataset compared to the current state of the art.

Elias Marks, Jonas Bömer, Federico Magistri et al.

Agricultural production is facing severe challenges in the next decades induced by climate change and the need for sustainability, reducing its impact on the environment. Advancements in field management through non-chemical weeding by robots in combination with monitoring of crops by autonomous unmanned aerial vehicles (UAVs) and breeding of novel and more resilient crop varieties are helpful to address these challenges. The analysis of plant traits, called phenotyping, is an essential activity in plant breeding, it however involves a great amount of manual labor. With this paper, we address the problem of automatic fine-grained organ-level geometric analysis needed for precision phenotyping. As the availability of real-world data in this domain is relatively scarce, we propose a novel dataset that was acquired using UAVs capturing high-resolution images of a real breeding trial containing 48 plant varieties and therefore covering great morphological and appearance diversity. This enables the development of approaches for autonomous phenotyping that generalize well to different varieties. Based on overlapping high-resolution images from multiple viewing angles, we compute photogrammetric dense point clouds and provide detailed and accurate point-wise labels for plants, leaves, and salient points as the tip and the base. Additionally, we include measurements of phenotypic traits performed by experts from the German Federal Plant Variety Office on the real plants, allowing the evaluation of new approaches not only on segmentation and keypoint detection but also directly on the downstream tasks. The provided labeled point clouds enable fine-grained plant analysis and support further progress in the development of automatic phenotyping approaches, but also enable further research in surface reconstruction, point cloud completion, and semantic interpretation of point clouds.

Matthias Zeller, Daniel Casado Herraez, Bengisu Ayan et al.

Semantic scene understanding, including the perception and classification of moving agents, is essential to enabling safe and robust driving behaviours of autonomous vehicles. Cameras and LiDARs are commonly used for semantic scene understanding. However, both sensor modalities face limitations in adverse weather and usually do not provide motion information. Radar sensors overcome these limitations and directly offer information about moving agents by measuring the Doppler velocity, but the measurements are comparably sparse and noisy. In this paper, we address the problem of panoptic segmentation in sparse radar point clouds to enhance scene understanding. Our approach, called SemRaFiner, accounts for changing density in sparse radar point clouds and optimizes the feature extraction to improve accuracy. Furthermore, we propose an optimized training procedure to refine instance assignments by incorporating a dedicated data augmentation. Our experiments suggest that our approach outperforms state-of-the-art methods for radar-based panoptic segmentation.

Christian Witte, Jens Behley, Cyrill Stachniss et al.

Spatial understanding of the semantics of the surroundings is a key capability needed by autonomous cars to enable safe driving decisions. Recently, purely vision-based solutions have gained increasing research interest. In particular, approaches extracting a bird's eye view (BEV) from multiple cameras have demonstrated great performance for spatial understanding. This paper addresses the dependency on learned positional encodings to correlate image and BEV feature map elements for transformer-based methods. We propose leveraging epipolar geometric constraints to model the relationship between cameras and the BEV by Epipolar Attention Fields. They are incorporated into the attention mechanism as a novel attribution term, serving as an alternative to learned positional encodings. Experiments show that our method EAFormer outperforms previous BEV approaches by 2% mIoU for map semantic segmentation and exhibits superior generalization capabilities compared to implicitly learning the camera configuration.

Daniel Fusaro, Federico Magistri, Jens Behley et al.

Accurate and consistent fruit monitoring over time is a key step toward automated agricultural production systems. However, this task is inherently difficult due to variations in fruit size, shape, occlusion, orientation, and the dynamic nature of orchards where fruits may appear or disappear between observations. In this article, we propose a novel method for fruit instance segmentation and re-identification on 3D terrestrial point clouds collected over time. Our approach directly operates on dense colored point clouds, capturing fine-grained 3D spatial detail. We segment individual fruits using a learning-based instance segmentation method applied directly to the point cloud. For each segmented fruit, we extract a compact and discriminative descriptor using a 3D sparse convolutional neural network. To track fruits across different times, we introduce an attention-based matching network that associates fruits with their counterparts from previous sessions. Matching is performed using a probabilistic assignment scheme, selecting the most likely associations across time. We evaluate our approach on real-world datasets of strawberries and apples, demonstrating that it outperforms existing methods in both instance segmentation and temporal re-identification, enabling robust and precise fruit monitoring across complex and dynamic orchard environments.

Khanh Son Pham, Christian Witte, Jens Behley et al.

Most autonomous cars rely on the availability of high-definition (HD) maps. Current research aims to address this constraint by directly predicting HD map elements from onboard sensors and reasoning about the relationships between the predicted map and traffic elements. Despite recent advancements, the coherent online construction of HD maps remains a challenging endeavor, as it necessitates modeling the high complexity of road topologies in a unified and consistent manner. To address this challenge, we propose a coherent approach to predict lane segments and their corresponding topology, as well as road boundaries, all by leveraging prior map information represented by commonly available standard-definition (SD) maps. We propose a network architecture, which leverages hybrid lane segment encodings comprising prior information and denoising techniques to enhance training stability and performance. Furthermore, we facilitate past frames for temporal consistency. Our experimental evaluation demonstrates that our approach outperforms previous methods by a large margin, highlighting the benefits of our modeling scheme.

Arun N. Sivakumar, Federico Magistri, Mateus V. Gasparino et al.

Under-canopy agricultural robots can enable various applications like precise monitoring, spraying, weeding, and plant manipulation tasks throughout the growing season. Autonomous navigation under the canopy is challenging due to the degradation in accuracy of RTK-GPS and the large variability in the visual appearance of the scene over time. In prior work, we developed a supervised learning-based perception system with semantic keypoint representation and deployed this in various field conditions. A large number of failures of this system can be attributed to the inability of the perception model to adapt to the domain shift encountered during deployment. In this paper, we propose a self-supervised online adaptation method for adapting the semantic keypoint representation using a visual foundational model, geometric prior, and pseudo labeling. Our preliminary experiments show that with minimal data and fine-tuning of parameters, the keypoint prediction model trained with labels on the source domain can be adapted in a self-supervised manner to various challenging target domains onboard the robot computer using our method. This can enable fully autonomous row-following capability in under-canopy robots across fields and crops without requiring human intervention.

Gianmarco Roggiolani, Federico Magistri, Tiziano Guadagnino et al.

Crops for food, feed, fiber, and fuel are key natural resources for our society. Monitoring plants and measuring their traits is an important task in agriculture often referred to as plant phenotyping. Traditionally, this task is done manually, which is time- and labor-intensive. Robots can automate phenotyping providing reproducible and high-frequency measurements. Today's perception systems use deep learning to interpret these measurements, but require a substantial amount of annotated data to work well. Obtaining such labels is challenging as it often requires background knowledge on the side of the labelers. This paper addresses the problem of reducing the labeling effort required to perform leaf instance segmentation on 3D point clouds, which is a first step toward phenotyping in 3D. Separating all leaves allows us to count them and compute relevant traits as their areas, lengths, and widths. We propose a novel self-supervised task-specific pre-training approach to initialize the backbone of a network for leaf instance segmentation. We also introduce a novel automatic postprocessing that considers the difficulty of correctly segmenting the points close to the stem, where all the leaves petiole overlap. The experiments presented in this paper suggest that our approach boosts the performance over all the investigated scenarios. We also evaluate the embeddings to assess the quality of the fully unsupervised approach and see a higher performance of our domain-specific postprocessing.

Xieyuanli Chen, Benedikt Mersch, Lucas Nunes et al.

Understanding the scene is key for autonomously navigating vehicles and the ability to segment the surroundings online into moving and non-moving objects is a central ingredient for this task. Often, deep learning-based methods are used to perform moving object segmentation (MOS). The performance of these networks, however, strongly depends on the diversity and amount of labeled training data, information that may be costly to obtain. In this paper, we propose an automatic data labeling pipeline for 3D LiDAR data to save the extensive manual labeling effort and to improve the performance of existing learning-based MOS systems by automatically generating labeled training data. Our proposed approach achieves this by processing the data offline in batches. It first exploits an occupancy-based dynamic object removal to detect possible dynamic objects coarsely. Second, it extracts segments among the proposals and tracks them using a Kalman filter. Based on the tracked trajectories, it labels the actually moving objects such as driving cars and pedestrians as moving. In contrast, the non-moving objects, e.g., parked cars, lamps, roads, or buildings, are labeled as static. We show that this approach allows us to label LiDAR data highly effectively and compare our results to those of other label generation methods. We also train a deep neural network with our auto-generated labels and achieve similar performance compared to the one trained with manual labels on the same data, and an even better performance when using additional datasets with labels generated by our approach. Furthermore, we evaluate our method on multiple datasets using different sensors and our experiments indicate that our method can generate labels in diverse environments.

Benedikt Mersch, Xieyuanli Chen, Jens Behley et al.

Exploiting past 3D LiDAR scans to predict future point clouds is a promising method for autonomous mobile systems to realize foresighted state estimation, collision avoidance, and planning. In this paper, we address the problem of predicting future 3D LiDAR point clouds given a sequence of past LiDAR scans. Estimating the future scene on the sensor level does not require any preceding steps as in localization or tracking systems and can be trained self-supervised. We propose an end-to-end approach that exploits a 2D range image representation of each 3D LiDAR scan and concatenates a sequence of range images to obtain a 3D tensor. Based on such tensors, we develop an encoder-decoder architecture using 3D convolutions to jointly aggregate spatial and temporal information of the scene and to predict the future 3D point clouds. We evaluate our method on multiple datasets and the experimental results suggest that our method outperforms existing point cloud prediction architectures and generalizes well to new, unseen environments without additional fine-tuning. Our method operates online and is faster than the common LiDAR frame rate of 10 Hz.

Xieyuanli Chen, Ignacio Vizzo, Thomas Läbe et al.

Robust and accurate, map-based localization is crucial for autonomous mobile systems. In this paper, we exploit range images generated from 3D LiDAR scans to address the problem of localizing mobile robots or autonomous cars in a map of a large-scale outdoor environment represented by a triangular mesh. We use the Poisson surface reconstruction to generate the mesh-based map representation. Based on the range images generated from the current LiDAR scan and the synthetic rendered views from the mesh-based map, we propose a new observation model and integrate it into a Monte Carlo localization framework, which achieves better localization performance and generalizes well to different environments. We test the proposed localization approach on multiple datasets collected in different environments with different LiDAR scanners. The experimental results show that our method can reliably and accurately localize a mobile system in different environments and operate online at the LiDAR sensor frame rate to track the vehicle pose.

Xieyuanli Chen, Thomas Läbe, Lorenzo Nardi et al.

Localization is a crucial capability for mobile robots and autonomous cars. In this paper, we address learning an observation model for Monte-Carlo localization using 3D LiDAR data. We propose a novel, neural network-based observation model that computes the expected overlap of two 3D LiDAR scans. The model predicts the overlap and yaw angle offset between the current sensor reading and virtual frames generated from a pre-built map. We integrate this observation model into a Monte-Carlo localization framework and tested it on urban datasets collected with a car in different seasons. The experiments presented in this paper illustrate that our method can reliably localize a vehicle in typical urban environments. We furthermore provide comparisons to a beam-end point and a histogram-based method indicating a superior global localization performance of our method with fewer particles.

Xieyuanli Chen, Thomas Läbe, Andres Milioto et al.

Simultaneous localization and mapping (SLAM) is a fundamental capability required by most autonomous systems. In this paper, we address the problem of loop closing for SLAM based on 3D laser scans recorded by autonomous cars. Our approach utilizes a deep neural network exploiting different cues generated from LiDAR data for finding loop closures. It estimates an image overlap generalized to range images and provides a relative yaw angle estimate between pairs of scans. Based on such predictions, we tackle loop closure detection and integrate our approach into an existing SLAM system to improve its mapping results. We evaluate our approach on sequences of the KITTI odometry benchmark and the Ford campus dataset. We show that our method can effectively detect loop closures surpassing the detection performance of state-of-the-art methods. To highlight the generalization capabilities of our approach, we evaluate our model on the Ford campus dataset while using only KITTI for training. The experiments show that the learned representation is able to provide reliable loop closure candidates, also in unseen environments.

Xieyuanli Chen, Andres Milioto, Emanuele Palazzolo et al.

Reliable and accurate localization and mapping are key components of most autonomous systems. Besides geometric information about the mapped environment, the semantics plays an important role to enable intelligent navigation behaviors. In most realistic environments, this task is particularly complicated due to dynamics caused by moving objects, which can corrupt the mapping step or derail localization. In this paper, we propose an extension of a recently published surfel-based mapping approach exploiting three-dimensional laser range scans by integrating semantic information to facilitate the mapping process. The semantic information is efficiently extracted by a fully convolutional neural network and rendered on a spherical projection of the laser range data. This computed semantic segmentation results in point-wise labels for the whole scan, allowing us to build a semantically-enriched map with labeled surfels. This semantic map enables us to reliably filter moving objects, but also improve the projective scan matching via semantic constraints. Our experimental evaluation on challenging highways sequences from KITTI dataset with very few static structures and a large amount of moving cars shows the advantage of our semantic SLAM approach in comparison to a purely geometric, state-of-the-art approach.

Xieyuanli Chen, Shijie Li, Benedikt Mersch et al.

The ability to detect and segment moving objects in a scene is essential for building consistent maps, making future state predictions, avoiding collisions, and planning. In this paper, we address the problem of moving object segmentation from 3D LiDAR scans. We propose a novel approach that pushes the current state of the art in LiDAR-only moving object segmentation forward to provide relevant information for autonomous robots and other vehicles. Instead of segmenting the point cloud semantically, i.e., predicting the semantic classes such as vehicles, pedestrians, roads, etc., our approach accurately segments the scene into moving and static objects, i.e., also distinguishing between moving cars vs. parked cars. Our proposed approach exploits sequential range images from a rotating 3D LiDAR sensor as an intermediate representation combined with a convolutional neural network and runs faster than the frame rate of the sensor. We compare our approach to several other state-of-the-art methods showing superior segmentation quality in urban environments. Additionally, we created a new benchmark for LiDAR-based moving object segmentation based on SemanticKITTI. We published it to allow other researchers to compare their approaches transparently and we furthermore published our code.

Mehmet Aygün, Aljoša Ošep, Mark Weber et al.

Temporal semantic scene understanding is critical for self-driving cars or robots operating in dynamic environments. In this paper, we propose 4D panoptic LiDAR segmentation to assign a semantic class and a temporally-consistent instance ID to a sequence of 3D points. To this end, we present an approach and a point-centric evaluation metric. Our approach determines a semantic class for every point while modeling object instances as probability distributions in the 4D spatio-temporal domain. We process multiple point clouds in parallel and resolve point-to-instance associations, effectively alleviating the need for explicit temporal data association. Inspired by recent advances in benchmarking of multi-object tracking, we propose to adopt a new evaluation metric that separates the semantic and point-to-instance association aspects of the task. With this work, we aim at paving the road for future developments of temporal LiDAR panoptic perception.

Nived Chebrolu, Thomas Läbe, Olga Vysotska et al.

State estimation is a key ingredient in most robotic systems. Often, state estimation is performed using some form of least squares minimization. Basically, all error minimization procedures that work on real-world data use robust kernels as the standard way for dealing with outliers in the data. These kernels, however, are often hand-picked, sometimes in different combinations, and their parameters need to be tuned manually for a particular problem. In this paper, we propose the use of a generalized robust kernel family, which is automatically tuned based on the distribution of the residuals and includes the common m-estimators. We tested our adaptive kernel with two popular estimation problems in robotics, namely ICP and bundle adjustment. The experiments presented in this paper suggest that our approach provides higher robustness while avoiding a manual tuning of the kernel parameters.

Jens Behley, Andres Milioto, Cyrill Stachniss

Panoptic segmentation is the recently introduced task that tackles semantic segmentation and instance segmentation jointly. In this paper, we present an extension of SemanticKITTI, which is a large-scale dataset providing dense point-wise semantic labels for all sequences of the KITTI Odometry Benchmark, for training and evaluation of laser-based panoptic segmentation. We provide the data and discuss the processing steps needed to enrich a given semantic annotation with temporally consistent instance information, i.e., instance information that supplements the semantic labels and identifies the same instance over sequences of LiDAR point clouds. Additionally, we present two strong baselines that combine state-of-the-art LiDAR-based semantic segmentation approaches with a state-of-the-art detector enriching the segmentation with instance information and that allow other researchers to compare their approaches against. We hope that our extension of SemanticKITTI with strong baselines enables the creation of novel algorithms for LiDAR-based panoptic segmentation as much as it has for the original semantic segmentation and semantic scene completion tasks. Data, code, and an online evaluation using a hidden test set will be published on http://semantic-kitti.org.

Jens Behley, Martin Garbade, Andres Milioto et al.

Semantic scene understanding is important for various applications. In particular, self-driving cars need a fine-grained understanding of the surfaces and objects in their vicinity. Light detection and ranging (LiDAR) provides precise geometric information about the environment and is thus a part of the sensor suites of almost all self-driving cars. Despite the relevance of semantic scene understanding for this application, there is a lack of a large dataset for this task which is based on an automotive LiDAR. In this paper, we introduce a large dataset to propel research on laser-based semantic segmentation. We annotated all sequences of the KITTI Vision Odometry Benchmark and provide dense point-wise annotations for the complete $360^{o}$ field-of-view of the employed automotive LiDAR. We propose three benchmark tasks based on this dataset: (i) semantic segmentation of point clouds using a single scan, (ii) semantic segmentation using multiple past scans, and (iii) semantic scene completion, which requires to anticipate the semantic scene in the future. We provide baseline experiments and show that there is a need for more sophisticated models to efficiently tackle these tasks. Our dataset opens the door for the development of more advanced methods, but also provides plentiful data to investigate new research directions.

Philipp Lottes, Jens Behley, Nived Chebrolu et al.

Applying agrochemicals is the default procedure for conventional weed control in crop production, but has negative impacts on the environment. Robots have the potential to treat every plant in the field individually and thus can reduce the required use of such chemicals. To achieve that, robots need the ability to identify crops and weeds in the field and must additionally select effective treatments. While certain types of weed can be treated mechanically, other types need to be treated by (selective) spraying. In this paper, we present an approach that provides the necessary information for effective plant-specific treatment. It outputs the stem location for weeds, which allows for mechanical treatments, and the covered area of the weed for selective spraying. Our approach uses an end-to-end trainable fully convolutional network that simultaneously estimates stem positions as well as the covered area of crops and weeds. It jointly learns the class-wise stem detection and the pixel-wise semantic segmentation. Experimental evaluations on different real-world datasets show that our approach is able to reliably solve this problem. Compared to state-of-the-art approaches, our approach not only substantially improves the stem detection accuracy, i.e., distinguishing crop and weed stems, but also provides an improvement in the semantic segmentation performance.

Philipp Lottes, Jens Behley, Andres Milioto et al.

Reducing the use of agrochemicals is an important component towards sustainable agriculture. Robots that can perform targeted weed control offer the potential to contribute to this goal, for example, through specialized weeding actions such as selective spraying or mechanical weed removal. A prerequisite of such systems is a reliable and robust plant classification system that is able to distinguish crop and weed in the field. A major challenge in this context is the fact that different fields show a large variability. Thus, classification systems have to robustly cope with substantial environmental changes with respect to weed pressure and weed types, growth stages of the crop, visual appearance, and soil conditions. In this paper, we propose a novel crop-weed classification system that relies on a fully convolutional network with an encoder-decoder structure and incorporates spatial information by considering image sequences. Exploiting the crop arrangement information that is observable from the image sequences enables our system to robustly estimate a pixel-wise labeling of the images into crop and weed, i.e., a semantic segmentation. We provide a thorough experimental evaluation, which shows that our system generalizes well to previously unseen fields under varying environmental conditions --- a key capability to actually use such systems in precision framing. We provide comparisons to other state-of-the-art approaches and show that our system substantially improves the accuracy of crop-weed classification without requiring a retraining of the model.