Patric Jensfelt

Qingwen Zhang, Yi Yang, Peizheng Li et al.

Scene flow estimation predicts the 3D motion at each point in successive LiDAR scans. This detailed, point-level, information can help autonomous vehicles to accurately predict and understand dynamic changes in their surroundings. Current state-of-the-art methods require annotated data to train scene flow networks and the expense of labeling inherently limits their scalability. Self-supervised approaches can overcome the above limitations, yet face two principal challenges that hinder optimal performance: point distribution imbalance and disregard for object-level motion constraints. In this paper, we propose SeFlow, a self-supervised method that integrates efficient dynamic classification into a learning-based scene flow pipeline. We demonstrate that classifying static and dynamic points helps design targeted objective functions for different motion patterns. We also emphasize the importance of internal cluster consistency and correct object point association to refine the scene flow estimation, in particular on object details. Our real-time capable method achieves state-of-the-art performance on the self-supervised scene flow task on Argoverse 2 and Waymo datasets. The code is open-sourced at https://github.com/KTH-RPL/SeFlow along with trained model weights.

Fereidoon Zangeneh, Leonard Bruns, Amit Dekel et al.

Visual localization allows autonomous robots to relocalize when losing track of their pose by matching their current observation with past ones. However, ambiguous scenes pose a challenge for such systems, as repetitive structures can be viewed from many distinct, equally likely camera poses, which means it is not sufficient to produce a single best pose hypothesis. In this work, we propose a probabilistic framework that for a given image predicts the arbitrarily shaped posterior distribution of its camera pose. We do this via a novel formulation of camera pose regression using variational inference, which allows sampling from the predicted distribution. Our method outperforms existing methods on localization in ambiguous scenes. Code and data will be released at https://github.com/efreidun/vapor.

Leonard Bruns, Patric Jensfelt

Rich geometric understanding of the world is an important component of many robotic applications such as planning and manipulation. In this paper, we present a modular pipeline for pose and shape estimation of objects from RGB-D images given their category. The core of our method is a generative shape model, which we integrate with a novel initialization network and a differentiable renderer to enable 6D pose and shape estimation from a single or multiple views. We investigate the use of discretized signed distance fields as an efficient shape representation for fast analysis-by-synthesis optimization. Our modular framework enables multi-view optimization and extensibility. We demonstrate the benefits of our approach over state-of-the-art methods in several experiments on both synthetic and real data. We open-source our approach at https://github.com/roym899/sdfest.

Leonard Bruns, Patric Jensfelt

Recently, various methods for 6D pose and shape estimation of objects at a per-category level have been proposed. This work provides an overview of the field in terms of methods, datasets, and evaluation protocols. First, an overview of existing works and their commonalities and differences is provided. Second, we take a critical look at the predominant evaluation protocol, including metrics and datasets. Based on the findings, we propose a new set of metrics, contribute new annotations for the Redwood dataset, and evaluate state-of-the-art methods in a fair comparison. The results indicate that existing methods do not generalize well to unconstrained orientations and are actually heavily biased towards objects being upright. We provide an easy-to-use evaluation toolbox with well-defined metrics, methods, and dataset interfaces, which allows evaluation and comparison with various state-of-the-art approaches (https://github.com/roym899/pose_and_shape_evaluation).

Qingwen Zhang, Chenhan Jiang, Xiaomeng Zhu et al.

Self-supervised feed-forward methods for scene flow estimation offer real-time efficiency, but their supervision from two-frame point correspondences is unreliable and often breaks down under occlusions. Multi-frame supervision has the potential to provide more stable guidance by incorporating motion cues from past frames, yet naive extensions of two-frame objectives are ineffective because point correspondences vary abruptly across frames, producing inconsistent signals. In the paper, we present TeFlow, enabling multi-frame supervision for feed-forward models by mining temporally consistent supervision. TeFlow introduces a temporal ensembling strategy that forms reliable supervisory signals by aggregating the most temporally consistent motion cues from a candidate pool built across multiple frames. Extensive evaluations demonstrate that TeFlow establishes a new state-of-the-art for self-supervised feed-forward methods, achieving performance gains of up to 33\% on the challenging Argoverse 2 and nuScenes datasets. Our method performs on par with leading optimization-based methods, yet speeds up 150 times. The code is open-sourced at https://github.com/KTH-RPL/OpenSceneFlow along with trained model weights.

Qingwen Zhang, Daniel Duberg, Ruoyu Geng et al.

In the field of robotics, the point cloud has become an essential map representation. From the perspective of downstream tasks like localization and global path planning, points corresponding to dynamic objects will adversely affect their performance. Existing methods for removing dynamic points in point clouds often lack clarity in comparative evaluations and comprehensive analysis. Therefore, we propose an easy-to-extend unified benchmarking framework for evaluating techniques for removing dynamic points in maps. It includes refactored state-of-art methods and novel metrics to analyze the limitations of these approaches. This enables researchers to dive deep into the underlying reasons behind these limitations. The benchmark makes use of several datasets with different sensor types. All the code and datasets related to our study are publicly available for further development and utilization.

Ajinkya Khoche, Laura Pereira Sánchez, Nazre Batool et al.

3D object detection at long range is crucial for ensuring the safety and efficiency of self driving vehicles, allowing them to accurately perceive and react to objects, obstacles, and potential hazards from a distance. But most current state of the art LiDAR based methods are range limited due to sparsity at long range, which generates a form of domain gap between points closer to and farther away from the ego vehicle. Another related problem is the label imbalance for faraway objects, which inhibits the performance of Deep Neural Networks at long range. To address the above limitations, we investigate two ways to improve long range performance of current LiDAR based 3D detectors. First, we combine two 3D detection networks, referred to as range experts, one specializing at near to mid range objects, and one at long range 3D detection. To train a detector at long range under a scarce label regime, we further weigh the loss according to the labelled point's distance from ego vehicle. Second, we augment LiDAR scans with virtual points generated using Multimodal Virtual Points (MVP), a readily available image-based depth completion algorithm. Our experiments on the long range Argoverse2 (AV2) dataset indicate that MVP is more effective in improving long range performance, while maintaining a straightforward implementation. On the other hand, the range experts offer a computationally efficient and simpler alternative, avoiding dependency on image-based segmentation networks and perfect camera-LiDAR calibration.

Lena Wild, Ludvig Ericson, Rafael Valencia et al.

Acquisition and maintenance are central problems in deploying high-definition (HD) maps for autonomous driving, with two lines of research prevalent in current literature: Online HD map generation and HD map change detection. However, the generated map's quality is currently insufficient for safe deployment, and many change detection approaches fail to precisely localize and extract the changed map elements, hence lacking explainability and hindering a potential fleet-based cooperative HD map update. In this paper, we propose the novel task of explainable element-based HD map change detection and update. In extending recent approaches that use online mapping techniques informed with an outdated map prior for HD map updating, we present ExelMap, an explainable element-based map updating strategy that specifically identifies changed map elements. In this context, we discuss how currently used metrics fail to capture change detection performance, while allowing for unfair comparison between prior-less and prior-informed map generation methods. Finally, we present an experimental study on real-world changes related to pedestrian crossings of the Argoverse 2 Map Change Dataset. To the best of our knowledge, this is the first comprehensive problem investigation of real-world end-to-end element-based HD map change detection and update, and ExelMap the first proposed solution.

Maciej K. Wozniak, Viktor Karefjards, Marko Thiel et al.

Multimodal sensor fusion methods for 3D object detection have been revolutionizing the autonomous driving research field. Nevertheless, most of these methods heavily rely on dense LiDAR data and accurately calibrated sensors which is often not the case in real-world scenarios. Data from LiDAR and cameras often come misaligned due to the miscalibration, decalibration, or different frequencies of the sensors. Additionally, some parts of the LiDAR data may be occluded and parts of the data may be missing due to hardware malfunction or weather conditions. This work presents a novel fusion step that addresses data corruptions and makes sensor fusion for 3D object detection more robust. Through extensive experiments, we demonstrate that our method performs on par with state-of-the-art approaches on normal data and outperforms them on misaligned data.

Ludvig Ericson, Patric Jensfelt

Floor plans are the basis of reasoning in and communicating about indoor environments. In this paper, we show that by modelling floor plans as sequences of line segments seen from a particular point of view, recent advances in autoregressive sequence modelling can be leveraged to model and predict floor plans. The line segments are canonicalized and translated to sequence of tokens and an attention-based neural network is used to fit a one-step distribution over next tokens. We fit the network to sequences derived from a set of large-scale floor plans, and demonstrate the capabilities of the model in four scenarios: novel floor plan generation, completion of partially observed floor plans, generation of floor plans from simulated sensor data, and finally, the applicability of a floor plan model in predicting the shortest distance with partial knowledge of the environment.

Bingxue Xu, Emil Hedemalm, Ajinkya Khoche et al.

The challenge of 3D multi-object tracking is achieving robustness in real-world applications, for example under adverse conditions and maintaining consistency as distance increases. To overcome these challenges, sensor fusion approaches that combine LiDAR, cameras, and radar have emerged. However, existing multimodal methods usually treat radar as another learned feature inside the network. When the overall model degrades in difficult environments, the robustness advantages that radar could provide are also reduced. In this paper we propose RadarMOT, a radar-informed 3D multi-object tracking framework that explicitly uses radar point clouds as additional observations to refine state estimation and recover objects missed by the detector at long ranges. Evaluations on the MAN-TruckScenes dataset show that RadarMOT consistently improves the Average Multi-Object Tracking Accuracy (AMOTA) by 12.7\% at long range and up to 10.3\% in adverse weather. The code will be available at https://github.com/bingxue-xu/radarmot

Leonard Bruns, Fereidoon Zangeneh, Patric Jensfelt

We consider the problem of tracking the 6D pose of a moving RGB-D camera in a neural scene representation. Different such representations have recently emerged, and we investigate the suitability of them for the task of camera tracking. In particular, we propose to track an RGB-D camera using a signed distance field-based representation and show that compared to density-based representations, tracking can be sped up, which enables more robust and accurate pose estimates when computation time is limited.

Qingwen Zhang, Yi Yang, Heng Fang et al.

Scene flow estimation determines a scene's 3D motion field, by predicting the motion of points in the scene, especially for aiding tasks in autonomous driving. Many networks with large-scale point clouds as input use voxelization to create a pseudo-image for real-time running. However, the voxelization process often results in the loss of point-specific features. This gives rise to a challenge in recovering those features for scene flow tasks. Our paper introduces DeFlow which enables a transition from voxel-based features to point features using Gated Recurrent Unit (GRU) refinement. To further enhance scene flow estimation performance, we formulate a novel loss function that accounts for the data imbalance between static and dynamic points. Evaluations on the Argoverse 2 scene flow task reveal that DeFlow achieves state-of-the-art results on large-scale point cloud data, demonstrating that our network has better performance and efficiency compared to others. The code is open-sourced at https://github.com/KTH-RPL/deflow.

Mingkai Jia, Qingwen Zhang, Bowen Yang et al.

Global point clouds that correctly represent the static environment features can facilitate accurate localization and robust path planning. However, dynamic objects introduce undesired ghost tracks that are mixed up with the static environment. Existing dynamic removal methods normally fail to balance the performance in computational efficiency and accuracy. In response, we present BeautyMap to efficiently remove the dynamic points while retaining static features for high-fidelity global maps. Our approach utilizes a binary-encoded matrix to efficiently extract the environment features. With a bit-wise comparison between matrices of each frame and the corresponding map region, we can extract potential dynamic regions. Then we use coarse to fine hierarchical segmentation of the $z$-axis to handle terrain variations. The final static restoration module accounts for the range-visibility of each single scan and protects static points out of sight. Comparative experiments underscore BeautyMap's superior performance in both accuracy and efficiency against other dynamic points removal methods. The code is open-sourced at https://github.com/MKJia/BeautyMap.

Maciej K Wozniak, Mattias Hansson, Marko Thiel et al.

In this study, we address a gap in existing unsupervised domain adaptation approaches on LiDAR-based 3D object detection, which have predominantly concentrated on adapting between established, high-density autonomous driving datasets. We focus on sparser point clouds, capturing scenarios from different perspectives: not just from vehicles on the road but also from mobile robots on sidewalks, which encounter significantly different environmental conditions and sensor configurations. We introduce Unsupervised Adversarial Domain Adaptation for 3D Object Detection (UADA3D). UADA3D does not depend on pre-trained source models or teacher-student architectures. Instead, it uses an adversarial approach to directly learn domain-invariant features. We demonstrate its efficacy in various adaptation scenarios, showing significant improvements in both self-driving car and mobile robot domains. Our code is open-source and will be available soon.

Ajinkya Khoche, Qingwen Zhang, Laura Pereira Sanchez et al.

Scene flow enables an understanding of the motion characteristics of the environment in the 3D world. It gains particular significance in the long-range, where object-based perception methods might fail due to sparse observations far away. Although significant advancements have been made in scene flow pipelines to handle large-scale point clouds, a gap remains in scalability with respect to long-range. We attribute this limitation to the common design choice of using dense feature grids, which scale quadratically with range. In this paper, we propose Sparse Scene Flow (SSF), a general pipeline for long-range scene flow, adopting a sparse convolution based backbone for feature extraction. This approach introduces a new challenge: a mismatch in size and ordering of sparse feature maps between time-sequential point scans. To address this, we propose a sparse feature fusion scheme, that augments the feature maps with virtual voxels at missing locations. Additionally, we propose a range-wise metric that implicitly gives greater importance to faraway points. Our method, SSF, achieves state-of-the-art results on the Argoverse2 dataset, demonstrating strong performance in long-range scene flow estimation. Our code will be released at https://github.com/KTH-RPL/SSF.git.

Maciej K. Wozniak, Lianhang Liu, Yixi Cai et al.

While end-to-end autonomous driving models show promising results, their practical deployment is often hindered by large model sizes, a reliance on expensive LiDAR sensors and computationally intensive BEV feature representations. This limits their scalability, especially for mass-market vehicles equipped only with cameras. To address these challenges, we propose PRIX (Plan from Raw Pixels). Our novel and efficient end-to-end driving architecture operates using only camera data, without explicit BEV representation and forgoing the need for LiDAR. PRIX leverages a visual feature extractor coupled with a generative planning head to predict safe trajectories from raw pixel inputs directly. A core component of our architecture is the Context-aware Recalibration Transformer (CaRT), a novel module designed to effectively enhance multi-level visual features for more robust planning. We demonstrate through comprehensive experiments that PRIX achieves state-of-the-art performance on the NavSim and nuScenes benchmarks, matching the capabilities of larger, multimodal diffusion planners while being significantly more efficient in terms of inference speed and model size, making it a practical solution for real-world deployment. Our work is open-source and the code will be at https://maxiuw.github.io/prix.

Qingwen Zhang, Xiaomeng Zhu, Chenhan Jiang et al.

Reliable 3D dynamic perception requires models that can anticipate motion beyond predefined categories, yet progress is hindered by the scarcity of dense, high-quality motion annotations. While self-supervision on unlabeled real data offers a path forward, empirical evidence suggests that scaling unlabeled data fails to close the performance gap due to noisy proxy signals. In this paper, we propose a shift in paradigm: learning robust real-world motion priors entirely from scalable simulation. We introduce SynFlow, a data generation pipeline that generates large-scale synthetic dataset specifically designed for LiDAR scene flow. Unlike prior works that prioritize sensor-specific realism, SynFlow employs a motion-oriented strategy to synthesize diverse kinematic patterns across 4,000 sequences ($\sim$940k frames), termed SynFlow-4k. This represents a 34x scale-up in annotated volume over existing real-world benchmarks. Our experiments demonstrate that SynFlow-4k provides a highly domain-invariant motion prior. In a zero-shot regime, models trained exclusively on our synthetic data generalize across multiple real-world benchmarks, rivaling in-domain supervised baselines on nuScenes and outperforming state-of-the-art methods on TruckScenes by 31.8%. Furthermore, SynFlow-4k serves as a label-efficient foundation: fine-tuning with only 5% of real-world labels surpasses models trained from scratch on the full available budget. We open-source the pipeline and dataset to facilitate research in generalizable 3D motion estimation. More detail can be found at https://kin-zhang.github.io/SynFlow.

Ajinkya Khoche, Gergő László Nagy, Maciej Wozniak et al.

Zero-shot 3D object classification is crucial for real-world applications like autonomous driving, however it is often hindered by a significant domain gap between the synthetic data used for training and the sparse, noisy LiDAR scans encountered in the real-world. Current methods trained solely on synthetic data fail to generalize to outdoor scenes, while those trained only on real data lack the semantic diversity to recognize rare or unseen objects. We introduce BlendCLIP, a multimodal pretraining framework that bridges this synthetic-to-real gap by strategically combining the strengths of both domains. We first propose a pipeline to generate a large-scale dataset of object-level triplets -- consisting of a point cloud, image, and text description -- mined directly from real-world driving data and human annotated 3D boxes. Our core contribution is a curriculum-based data mixing strategy that first grounds the model in the semantically rich synthetic CAD data before progressively adapting it to the specific characteristics of real-world scans. Our experiments show that our approach is highly label-efficient: introducing as few as 1.5\% real-world samples per batch into training boosts zero-shot accuracy on the nuScenes benchmark by 27\%. Consequently, our final model achieves state-of-the-art performance on challenging outdoor datasets like nuScenes and TruckScenes, improving over the best prior method by 19.3\% on nuScenes, while maintaining strong generalization on diverse synthetic benchmarks. Our findings demonstrate that effective domain adaptation, not full-scale real-world annotation, is the key to unlocking robust open-vocabulary 3D perception. Our code and dataset will be released upon acceptance on https://github.com/kesu1/BlendCLIP.

Qingwen Zhang, Xiaomeng Zhu, Yushan Zhang et al.

Previous dominant methods for scene flow estimation focus mainly on input from two consecutive frames, neglecting valuable information in the temporal domain. While recent trends shift towards multi-frame reasoning, they suffer from rapidly escalating computational costs as the number of frames grows. To leverage temporal information more efficiently, we propose DeltaFlow ($Δ$Flow), a lightweight 3D framework that captures motion cues via a $Δ$ scheme, extracting temporal features with minimal computational cost, regardless of the number of frames. Additionally, scene flow estimation faces challenges such as imbalanced object class distributions and motion inconsistency. To tackle these issues, we introduce a Category-Balanced Loss to enhance learning across underrepresented classes and an Instance Consistency Loss to enforce coherent object motion, improving flow accuracy. Extensive evaluations on the Argoverse 2, Waymo and nuScenes datasets show that $Δ$Flow achieves state-of-the-art performance with up to 22% lower error and $2\times$ faster inference compared to the next-best multi-frame supervised method, while also demonstrating a strong cross-domain generalization ability. The code is open-sourced at https://github.com/Kin-Zhang/DeltaFlow along with trained model weights.

Leonard Bruns, Jun Zhang, Patric Jensfelt

Neural field-based SLAM methods typically employ a single, monolithic field as their scene representation. This prevents efficient incorporation of loop closure constraints and limits scalability. To address these shortcomings, we propose a novel RGB-D neural mapping framework in which the scene is represented by a collection of lightweight neural fields which are dynamically anchored to the pose graph of a sparse visual SLAM system. Our approach shows the ability to integrate large-scale loop closures, while requiring only minimal reintegration. Furthermore, we verify the scalability of our approach by demonstrating successful building-scale mapping taking multiple loop closures into account during the optimization, and show that our method outperforms existing state-of-the-art approaches on large scenes in terms of quality and runtime. Our code is available open-source at https://github.com/KTH-RPL/neural_graph_mapping.

Leonard Bruns, Patric Jensfelt

Recently, various methods for 6D pose and shape estimation of objects have been proposed. Typically, these methods evaluate their pose estimation in terms of average precision, and reconstruction quality with chamfer distance. In this work we take a critical look at this predominant evaluation protocol including metrics and datasets. We propose a new set of metrics, contribute new annotations for the Redwood dataset and evaluate state-of-the-art methods in a fair comparison. We find that existing methods do not generalize well to unconstrained orientations, and are actually heavily biased towards objects being upright. We contribute an easy-to-use evaluation toolbox with well-defined metrics, method and dataset interfaces, which readily allows evaluation and comparison with various state-of-the-art approaches (see https://github.com/roym899/pose_and_shape_evaluation ).

Thien-Minh Nguyen, Shenghai Yuan, Thien Hoang Nguyen et al.

Perception plays a crucial role in various robot applications. However, existing well-annotated datasets are biased towards autonomous driving scenarios, while unlabelled SLAM datasets are quickly over-fitted, and often lack environment and domain variations. To expand the frontier of these fields, we introduce a comprehensive dataset named MCD (Multi-Campus Dataset), featuring a wide range of sensing modalities, high-accuracy ground truth, and diverse challenging environments across three Eurasian university campuses. MCD comprises both CCS (Classical Cylindrical Spinning) and NRE (Non-Repetitive Epicyclic) lidars, high-quality IMUs (Inertial Measurement Units), cameras, and UWB (Ultra-WideBand) sensors. Furthermore, in a pioneering effort, we introduce semantic annotations of 29 classes over 59k sparse NRE lidar scans across three domains, thus providing a novel challenge to existing semantic segmentation research upon this largely unexplored lidar modality. Finally, we propose, for the first time to the best of our knowledge, continuous-time ground truth based on optimization-based registration of lidar-inertial data on large survey-grade prior maps, which are also publicly released, each several times the size of existing ones. We conduct a rigorous evaluation of numerous state-of-the-art algorithms on MCD, report their performance, and highlight the challenges awaiting solutions from the research community.

Daniel Duberg, Qingwen Zhang, MingKai Jia et al.

The dynamic nature of the real world is one of the main challenges in robotics. The first step in dealing with it is to detect which parts of the world are dynamic. A typical benchmark task is to create a map that contains only the static part of the world to support, for example, localization and planning. Current solutions are often applied in post-processing, where parameter tuning allows the user to adjust the setting for a specific dataset. In this paper, we propose DUFOMap, a novel dynamic awareness mapping framework designed for efficient online processing. Despite having the same parameter settings for all scenarios, it performs better or is on par with state-of-the-art methods. Ray casting is utilized to identify and classify fully observed empty regions. Since these regions have been observed empty, it follows that anything inside them at another time must be dynamic. Evaluation is carried out in various scenarios, including outdoor environments in KITTI and Argoverse 2, open areas on the KTH campus, and with different sensor types. DUFOMap outperforms the state of the art in terms of accuracy and computational efficiency. The source code, benchmarks, and links to the datasets utilized are provided. See https://kth-rpl.github.io/dufomap for more details.

Ludvig Ericson, José Pedro, Patric Jensfelt

Autonomous exploration in mobile robotics often involves a trade-off between two objectives: maximizing environmental coverage and minimizing the total path length. In the widely used information gain paradigm, exploration is guided by the expected value of observations. While this approach is effective under budget-constrained settings--where only a limited number of observations can be made--it fails to align with quality-constrained scenarios, in which the robot must fully explore the environment to a desired level of certainty or quality. In such cases, total information gain is effectively fixed, and maximizing it per step can lead to inefficient, greedy behavior and unnecessary backtracking. This paper argues that information gain should not serve as an optimization objective in quality-constrained exploration. Instead, it should be used to filter viable candidate actions. We propose a novel heuristic, distance advantage, which selects candidate frontiers based on a trade-off between proximity to the robot and remoteness from other frontiers. This heuristic aims to reduce future detours by prioritizing exploration of isolated regions before the robot's opportunity to visit them efficiently has passed. We evaluate our method in simulated environments against classical frontier-based exploration and gain-maximizing approaches. Results show that distance advantage significantly reduces total path length across a variety of environments, both with and without access to prior map predictions. Our findings challenge the assumption that more accurate gain estimation improves performance and offer a more suitable alternative for the quality-constrained exploration paradigm.

Qingwen Zhang, Ajinkya Khoche, Yi Yang et al.

LiDAR point cloud is essential for autonomous vehicles, but motion distortions from dynamic objects degrade the data quality. While previous work has considered distortions caused by ego motion, distortions caused by other moving objects remain largely overlooked, leading to errors in object shape and position. This distortion is particularly pronounced in high-speed environments such as highways and in multi-LiDAR configurations, a common setup for heavy vehicles. To address this challenge, we introduce HiMo, a pipeline that repurposes scene flow estimation for non-ego motion compensation, correcting the representation of dynamic objects in point clouds. During the development of HiMo, we observed that existing self-supervised scene flow estimators often produce degenerate or inconsistent estimates under high-speed distortion. We further propose SeFlow++, a real-time scene flow estimator that achieves state-of-the-art performance on both scene flow and motion compensation. Since well-established motion distortion metrics are absent in the literature, we introduce two evaluation metrics: compensation accuracy at a point level and shape similarity of objects. We validate HiMo through extensive experiments on Argoverse 2, ZOD, and a newly collected real-world dataset featuring highway driving and multi-LiDAR-equipped heavy vehicles. Our findings show that HiMo improves the geometric consistency and visual fidelity of dynamic objects in LiDAR point clouds, benefiting downstream tasks such as semantic segmentation and 3D detection. See https://kin-zhang.github.io/HiMo for more details.

Ajinkya Khoche, Qingwen Zhang, Yixi Cai et al.

Accurate 3D scene flow estimation is critical for autonomous systems to navigate dynamic environments safely, but creating the necessary large-scale, manually annotated datasets remains a significant bottleneck for developing robust perception models. Current self-supervised methods struggle to match the performance of fully supervised approaches, especially in challenging long-range and adverse weather scenarios, while supervised methods are not scalable due to their reliance on expensive human labeling. We introduce DoGFlow, a novel self-supervised framework that recovers full 3D object motions for LiDAR scene flow estimation without requiring any manual ground truth annotations. This paper presents our cross-modal label transfer approach, where DoGFlow computes motion pseudo-labels in real-time directly from 4D radar Doppler measurements and transfers them to the LiDAR domain using dynamic-aware association and ambiguity-resolved propagation. On the challenging MAN TruckScenes dataset, DoGFlow substantially outperforms existing self-supervised methods and improves label efficiency by enabling LiDAR backbones to achieve over 90% of fully supervised performance with only 10% of the ground truth data. For more details, please visit https://ajinkyakhoche.github.io/DogFlow/

Ajinkya Khoche, Aron Asefaw, Alejandro Gonzalez et al.

Data annotation in autonomous vehicles is a critical step in the development of Deep Neural Network (DNN) based models or the performance evaluation of the perception system. This often takes the form of adding 3D bounding boxes on time-sequential and registered series of point-sets captured from active sensors like Light Detection and Ranging (LiDAR) and Radio Detection and Ranging (RADAR). When annotating multiple active sensors, there is a need to motion compensate and translate the points to a consistent coordinate frame and timestamp respectively. However, highly dynamic objects pose a unique challenge, as they can appear at different timestamps in each sensor's data. Without knowing the speed of the objects, their position appears to be different in different sensor outputs. Thus, even after motion compensation, highly dynamic objects are not matched from multiple sensors in the same frame, and human annotators struggle to add unique bounding boxes that capture all objects. This article focuses on addressing this challenge, primarily within the context of Scania collected datasets. The proposed solution takes a track of an annotated object as input and uses the Moving Horizon Estimation (MHE) to robustly estimate its speed. The estimated speed profile is utilized to correct the position of the annotated box and add boxes to object clusters missed by the original annotation.

Fereidoon Zangeneh, Leonard Bruns, Amit Dekel et al.

Accurate camera pose estimation from an image observation in a previously mapped environment is commonly done through structure-based methods: by finding correspondences between 2D keypoints on the image and 3D structure points in the map. In order to make this correspondence search tractable in large scenes, existing pipelines either rely on search heuristics, or perform image retrieval to reduce the search space by comparing the current image to a database of past observations. However, these approaches result in elaborate pipelines or storage requirements that grow with the number of past observations. In this work, we propose a new paradigm for making structure-based relocalisation tractable. Instead of relying on image retrieval or search heuristics, we learn a direct mapping from image observations to the visible scene structure in a compact neural network. Given a query image, a forward pass through our novel visible structure retrieval network allows obtaining the subset of 3D structure points in the map that the image views, thus reducing the search space of 2D-3D correspondences. We show that our proposed method enables performing localisation with an accuracy comparable to the state of the art, while requiring lower computational and storage footprint.

Ermanno Bartoli, Dennis Rotondi, Buwei He et al.

Understanding how people interact with their surroundings and each other is essential for enabling robots to act in socially compliant and context-aware ways. While 3D Scene Graphs have emerged as a powerful semantic representation for scene understanding, existing approaches largely ignore humans in the scene, also due to the lack of annotated human-environment relationships. Moreover, existing methods typically capture only open-vocabulary relations from single image frames, which limits their ability to model long-range interactions beyond the observed content. We introduce Social 3D Scene Graphs, an augmented 3D Scene Graph representation that captures humans, their attributes, activities and relationships in the environment, both local and remote, using an open-vocabulary framework. Furthermore, we introduce a new benchmark consisting of synthetic environments with comprehensive human-scene relationship annotations and diverse types of queries for evaluating social scene understanding in 3D. The experiments demonstrate that our representation improves human activity prediction and reasoning about human-environment relations, paving the way toward socially intelligent robots.

Lena Wild, Rafael Valencia, Patric Jensfelt

Reliable integration of prior information is crucial for self-verifying and self-updating HD maps. However, no public dataset includes the required triplet of prior maps, current maps, and sensor data. As a result, existing methods must rely on synthetic priors, which create inconsistencies and lead to a significant sim2real gap. To address this, we introduce ArgoTweak, the first dataset to complete the triplet with realistic map priors. At its core, ArgoTweak employs a bijective mapping framework, breaking down large-scale modifications into fine-grained atomic changes at the map element level, thus ensuring interpretability. This paradigm shift enables accurate change detection and integration while preserving unchanged elements with high fidelity. Experiments show that training models on ArgoTweak significantly reduces the sim2real gap compared to synthetic priors. Extensive ablations further highlight the impact of structured priors and detailed change annotations. By establishing a benchmark for explainable, prior-aided HD mapping, ArgoTweak advances scalable, self-improving mapping solutions. The dataset, baselines, map modification toolbox, and further resources are available at https://kth-rpl.github.io/ArgoTweak/.

Paula Carbó Cubero, Alberto Jaenal Gálvez, André Mateus et al.

State-of-the-art methods fail to solve visual localization in scenarios where different devices use different sparse feature extraction algorithms to obtain keypoints and their corresponding descriptors. Translating feature descriptors is enough to enable matching. However, performance is drastically reduced in cross-feature detector cases, because current solutions assume common keypoints. This means that the same detector has to be used, which is rarely the case in practice when different descriptors are used. The low repeatability of keypoints, in addition to non-discriminatory and non-distinctive descriptors, make the identification of true correspondences extremely challenging. We present the first method tackling this problem, which performs feature descriptor augmentation targeting cross-detector feature matching, and then feature translation to a latent space. We show that our method significantly improves image matching and visual localization in the cross-feature scenario and evaluate the proposed method on several benchmarks.

Fereidoon Zangeneh, Amit Dekel, Alessandro Pieropan et al.

Visual relocalization is the task of estimating the camera pose given an image it views. Absolute pose regression offers a solution to this task by training a neural network, directly regressing the camera pose from image features. While an attractive solution in terms of memory and compute efficiency, absolute pose regression's predictions are inaccurate and unreliable outside the training domain. In this work, we propose a novel method for quantifying the epistemic uncertainty of an absolute pose regression model by estimating the likelihood of observations within a variational framework. Beyond providing a measure of confidence in predictions, our approach offers a unified model that also handles observation ambiguities, probabilistically localizing the camera in the presence of repetitive structures. Our method outperforms existing approaches in capturing the relation between uncertainty and prediction error.

Ludvig Ericson, Patric Jensfelt

In this paper, we tackle the challenge of predicting the unseen walls of a partially observed environment as a set of 2D line segments, conditioned on occupancy grids integrated along the trajectory of a 360° LIDAR sensor. A dataset of such occupancy grids and their corresponding target wall segments is collected by navigating a virtual robot between a set of randomly sampled waypoints in a collection of office-scale floor plans from a university campus. The line segment prediction task is formulated as an autoregressive sequence prediction task, and an attention-based deep network is trained on the dataset. The sequence-based autoregressive formulation is evaluated through predicted information gain, as in frontier-based autonomous exploration, demonstrating significant improvements over both non-predictive estimation and convolution-based image prediction found in the literature. Ablations on key components are evaluated, as well as sensor range and the occupancy grid's metric area. Finally, model generality is validated by predicting walls in a novel floor plan reconstructed on-the-fly in a real-world office environment.

Daniel Duberg, Patric Jensfelt

3D models are an essential part of many robotic applications. In applications where the environment is unknown a-priori, or where only a part of the environment is known, it is important that the 3D model can handle the unknown space efficiently. Path planning, exploration, and reconstruction all fall into this category. In this paper we present an extension to OctoMap which we call UFOMap. UFOMap uses an explicit representation of all three states in the map, i.e., occupied, free, and unknown. This gives, surprisingly, a more memory efficient representation. Furthermore, we provide methods that allow for significantly faster insertions into the octree. This enables real-time colored volumetric mapping at high resolution (below 1 cm). UFOMap is contributed as a C++ library that can be used standalone but is also integrated into ROS.

Jiexiong Tang, Rares Ambrus, Vitor Guizilini et al.

Detecting and matching robust viewpoint-invariant keypoints is critical for visual SLAM and Structure-from-Motion. State-of-the-art learning-based methods generate training samples via homography adaptation to create 2D synthetic views with known keypoint matches from a single image. This approach, however, does not generalize to non-planar 3D scenes with illumination variations commonly seen in real-world videos. In this work, we propose self-supervised learning of depth-aware keypoints directly from unlabeled videos. We jointly learn keypoint and depth estimation networks by combining appearance and geometric matching via a differentiable structure-from-motion module based on Procrustean residual pose correction. We describe how our self-supervised keypoints can be integrated into state-of-the-art visual odometry frameworks for robust and accurate ego-motion estimation of autonomous vehicles in real-world conditions.

Tobias Klamt, Diego Rodriguez, Lorenzo Baccelliere et al.

Mobile manipulation robots have high potential to support rescue forces in disaster-response missions. Despite the difficulties imposed by real-world scenarios, robots are promising to perform mission tasks from a safe distance. In the CENTAURO project, we developed a disaster-response system which consists of the highly flexible Centauro robot and suitable control interfaces including an immersive tele-presence suit and support-operator controls on different levels of autonomy. In this article, we give an overview of the final CENTAURO system. In particular, we explain several high-level design decisions and how those were derived from requirements and extensive experience of Kerntechnische Hilfsdienst GmbH, Karlsruhe, Germany (KHG). We focus on components which were recently integrated and report about a systematic evaluation which demonstrated system capabilities and revealed valuable insights.

Xi Chen, Ali Ghadirzadeh, Mårten Björkman et al.

Deep reinforcement learning (RL) has enabled training action-selection policies, end-to-end, by learning a function which maps image pixels to action outputs. However, it's application to visuomotor robotic policy training has been limited because of the challenge of large-scale data collection when working with physical hardware. A suitable visuomotor policy should perform well not just for the task-setup it has been trained for, but also for all varieties of the task, including novel objects at different viewpoints surrounded by task-irrelevant objects. However, it is impractical for a robotic setup to sufficiently collect interactive samples in a RL framework to generalize well to novel aspects of a task. In this work, we demonstrate that by using adversarial training for domain transfer, it is possible to train visuomotor policies based on RL frameworks, and then transfer the acquired policy to other novel task domains. We propose to leverage the deep RL capabilities to learn complex visuomotor skills for uncomplicated task setups, and then exploit transfer learning to generalize to new task domains provided only still images of the task in the target domain. We evaluate our method on two real robotic tasks, picking and pouring, and compare it to a number of prior works, demonstrating its superiority.

Massimiliano Mancini, Hakan Karaoguz, Elisa Ricci et al.

While today's robots are able to perform sophisticated tasks, they can only act on objects they have been trained to recognize. This is a severe limitation: any robot will inevitably see new objects in unconstrained settings, and thus will always have visual knowledge gaps. However, standard visual modules are usually built on a limited set of classes and are based on the strong prior that an object must belong to one of those classes. Identifying whether an instance does not belong to the set of known categories (i.e. open set recognition), only partially tackles this problem, as a truly autonomous agent should be able not only to detect what it does not know, but also to extend dynamically its knowledge about the world. We contribute to this challenge with a deep learning architecture that can dynamically update its known classes in an end-to-end fashion. The proposed deep network, based on a deep extension of a non-parametric model, detects whether a perceived object belongs to the set of categories known by the system and learns it without the need to retrain the whole system from scratch. Annotated images about the new category can be provided by an 'oracle' (i.e. human supervision), or by autonomous mining of the Web. Experiments on two different databases and on a robot platform demonstrate the promise of our approach.

Jiexiong Tang, John Folkesson, Patric Jensfelt

In this paper, we proposed a new deep learning based dense monocular SLAM method. Compared to existing methods, the proposed framework constructs a dense 3D model via a sparse to dense mapping using learned surface normals. With single view learned depth estimation as prior for monocular visual odometry, we obtain both accurate positioning and high quality depth reconstruction. The depth and normal are predicted by a single network trained in a tightly coupled manner.Experimental results show that our method significantly improves the performance of visual tracking and depth prediction in comparison to the state-of-the-art in deep monocular dense SLAM.

Jiexiong Tang, Ludvig Ericson, John Folkesson et al.

In this paper, we present a deep learning-based network, GCNv2, for generation of keypoints and descriptors. GCNv2 is built on our previous method, GCN, a network trained for 3D projective geometry. GCNv2 is designed with a binary descriptor vector as the ORB feature so that it can easily replace ORB in systems such as ORB-SLAM2. GCNv2 significantly improves the computational efficiency over GCN that was only able to run on desktop hardware. We show how a modified version of ORB-SLAM2 using GCNv2 features runs on a Jetson TX2, an embedded low-power platform. Experimental results show that GCNv2 retains comparable accuracy as GCN and that it is robust enough to use for control of a flying drone.

Xi Chen, Ali Ghadirzadeh, Mårten Björkman et al.

Multi-objective reinforcement learning (MORL) is the generalization of standard reinforcement learning (RL) approaches to solve sequential decision making problems that consist of several, possibly conflicting, objectives. Generally, in such formulations, there is no single optimal policy which optimizes all the objectives simultaneously, and instead, a number of policies has to be found each optimizing a preference of the objectives. In other words, the MORL is framed as a meta-learning problem, with the task distribution given by a distribution over the preferences. We demonstrate that such a formulation results in a better approximation of the Pareto optimal solutions in terms of both the optimality and the computational efficiency. We evaluated our method on obtaining Pareto optimal policies using a number of continuous control problems with high degrees of freedom.

Massimiliano Mancini, Hakan Karaoguz, Elisa Ricci et al.

Technological developments call for increasing perception and action capabilities of robots. Among other skills, vision systems that can adapt to any possible change in the working conditions are needed. Since these conditions are unpredictable, we need benchmarks which allow to assess the generalization and robustness capabilities of our visual recognition algorithms. In this work we focus on robotic kitting in unconstrained scenarios. As a first contribution, we present a new visual dataset for the kitting task. Differently from standard object recognition datasets, we provide images of the same objects acquired under various conditions where camera, illumination and background are changed. This novel dataset allows for testing the robustness of robot visual recognition algorithms to a series of different domain shifts both in isolation and unified. Our second contribution is a novel online adaptation algorithm for deep models, based on batch-normalization layers, which allows to continuously adapt a model to the current working conditions. Differently from standard domain adaptation algorithms, it does not require any image from the target domain at training time. We benchmark the performance of the algorithm on the proposed dataset, showing its capability to fill the gap between the performances of a standard architecture and its counterpart adapted offline to the given target domain.

Xi Chen, Ali Ghadirzadeh, John Folkesson et al.

Mobile robot navigation in complex and dynamic environments is a challenging but important problem. Reinforcement learning approaches fail to solve these tasks efficiently due to reward sparsities, temporal complexities and high-dimensionality of sensorimotor spaces which are inherent in such problems. We present a novel approach to train action policies to acquire navigation skills for wheel-legged robots using deep reinforcement learning. The policy maps height-map image observations to motor commands to navigate to a target position while avoiding obstacles. We propose to acquire the multifaceted navigation skill by learning and exploiting a number of manageable navigation behaviors. We also introduce a domain randomization technique to improve the versatility of the training samples. We demonstrate experimentally a significant improvement in terms of data-efficiency, success rate, robustness against irrelevant sensory data, and also the quality of the maneuver skills.

Hakan Karaoguz, Patric Jensfelt

In this paper we address the problem of unsupervised localization of objects in single images. Compared to previous state-of-the-art method our method is fully unsupervised in the sense that there is no prior instance level or category level information about the image. Furthermore, we treat each image individually and do not rely on any neighboring image similarity. We employ deep-learning based generation of saliency maps and region proposals to tackle this problem. First salient regions in the image are determined using an encoder/decoder architecture. The resulting saliency map is matched with region proposals from a class agnostic region proposal network to roughly localize the candidate object regions. These regions are further refined based on the overlap and similarity ratios. Our experimental evaluations on a benchmark dataset show that the method gets close to current state-of-the-art methods in terms of localization accuracy even though these make use of multiple frames. Furthermore, we created a more challenging and realistic dataset with multiple object categories and varying viewpoint and illumination conditions for evaluating the method's performance in real world scenarios.

Nils Bore, Patric Jensfelt, John Folkesson

In this work, we present a method for tracking and learning the dynamics of all objects in a large scale robot environment. A mobile robot patrols the environment and visits the different locations one by one. Movable objects are discovered by change detection, and tracked throughout the robot deployment. For tracking, we extend the Rao-Blackwellized particle filter of previous work with birth and death processes, enabling the method to handle an arbitrary number of objects. Target births and associations are sampled using Gibbs sampling. The parameters of the system are then learnt using the Expectation Maximization algorithm in an unsupervised fashion. The system therefore enables learning of the dynamics of one particular environment, and of its objects. The algorithm is evaluated on data collected autonomously by a mobile robot in an office environment during a real-world deployment. We show that the algorithm automatically identifies and tracks the moving objects within 3D maps and infers plausible dynamics models, significantly decreasing the modeling bias of our previous work. The proposed method represents an improvement over previous methods for environment dynamics learning as it allows for learning of fine grained processes.

Nils Bore, Johan Ekekrantz, Patric Jensfelt et al.

This paper studies the problem of detection and tracking of general objects with long-term dynamics, observed by a mobile robot moving in a large environment. A key problem is that due to the environment scale, it can only observe a subset of the objects at any given time. Since some time passes between observations of objects in different places, the objects might be moved when the robot is not there. We propose a model for this movement in which the objects typically only move locally, but with some small probability they jump longer distances, through what we call global motion. For filtering, we decompose the posterior over local and global movements into two linked processes. The posterior over the global movements and measurement associations is sampled, while we track the local movement analytically using Kalman filters. This novel filter is evaluated on point cloud data gathered autonomously by a mobile robot over an extended period of time. We show that tracking jumping objects is feasible, and that the proposed probabilistic treatment outperforms previous methods when applied to real world data. The key to efficient probabilistic tracking in this scenario is focused sampling of the object posteriors.

Johan Ekekrantz, Nils Bore, Rares Ambrus et al.

In this paper we introduce a system for unsupervised object discovery and segmentation of RGBD-images. The system models the sensor noise directly from data, allowing accurate segmentation without sensor specific hand tuning of measurement noise models making use of the recently introduced Statistical Inlier Estimation (SIE) method. Through a fully probabilistic formulation, the system is able to apply probabilistic inference, enabling reliable segmentation in previously challenging scenarios. In addition, we introduce new methods for filtering out false positives, significantly improving the signal to noise ratio. We show that the system significantly outperform state-of-the-art in on a challenging real-world dataset.

Johan Ekekrantz, John Folkesson, Patric Jensfelt

In this paper we introduce an adaptive cost function for pointcloud registration. The algorithm automatically estimates the sensor noise, which is important for generalization across different sensors and environments. Through experiments on real and synthetic data, we show significant improvements in accuracy and robustness over state-of-the-art solutions.

Nick Hawes, Chris Burbridge, Ferdian Jovan et al.

Thanks to the efforts of the robotics and autonomous systems community, robots are becoming ever more capable. There is also an increasing demand from end-users for autonomous service robots that can operate in real environments for extended periods. In the STRANDS project we are tackling this demand head-on by integrating state-of-the-art artificial intelligence and robotics research into mobile service robots, and deploying these systems for long-term installations in security and care environments. Over four deployments, our robots have been operational for a combined duration of 104 days autonomously performing end-user defined tasks, covering 116km in the process. In this article we describe the approach we have used to enable long-term autonomous operation in everyday environments, and how our robots are able to use their long run times to improve their own performance.