Horst–Michael Groß

Daniel Seichter, Söhnke Benedikt Fischedick, Mona Köhler et al.

Semantic scene understanding is essential for mobile agents acting in various environments. Although semantic segmentation already provides a lot of information, details about individual objects as well as the general scene are missing but required for many real-world applications. However, solving multiple tasks separately is expensive and cannot be accomplished in real time given limited computing and battery capabilities on a mobile platform. In this paper, we propose an efficient multi-task approach for RGB-D scene analysis~(EMSANet) that simultaneously performs semantic and instance segmentation~(panoptic segmentation), instance orientation estimation, and scene classification. We show that all tasks can be accomplished using a single neural network in real time on a mobile platform without diminishing performance - by contrast, the individual tasks are able to benefit from each other. In order to evaluate our multi-task approach, we extend the annotations of the common RGB-D indoor datasets NYUv2 and SUNRGB-D for instance segmentation and orientation estimation. To the best of our knowledge, we are the first to provide results in such a comprehensive multi-task setting for indoor scene analysis on NYUv2 and SUNRGB-D.

Dustin Aganian, Benedict Stephan, Markus Eisenbach et al.

With the emergence of collaborative robots (cobots), human-robot collaboration in industrial manufacturing is coming into focus. For a cobot to act autonomously and as an assistant, it must understand human actions during assembly. To effectively train models for this task, a dataset containing suitable assembly actions in a realistic setting is crucial. For this purpose, we present the ATTACH dataset, which contains 51.6 hours of assembly with 95.2k annotated fine-grained actions monitored by three cameras, which represent potential viewpoints of a cobot. Since in an assembly context workers tend to perform different actions simultaneously with their two hands, we annotated the performed actions for each hand separately. Therefore, in the ATTACH dataset, more than 68% of annotations overlap with other annotations, which is many times more than in related datasets, typically featuring more simplistic assembly tasks. For better generalization with respect to the background of the working area, we did not only record color and depth images, but also used the Azure Kinect body tracking SDK for estimating 3D skeletons of the worker. To create a first baseline, we report the performance of state-of-the-art methods for action recognition as well as action detection on video and skeleton-sequence inputs. The dataset is available at https://www.tu-ilmenau.de/neurob/data-sets-code/attach-dataset .

Söhnke Benedikt Fischedick, Daniel Seichter, Robin Schmidt et al.

Scene analysis is essential for enabling autonomous systems, such as mobile robots, to operate in real-world environments. However, obtaining a comprehensive understanding of the scene requires solving multiple tasks, such as panoptic segmentation, instance orientation estimation, and scene classification. Solving these tasks given limited computing and battery capabilities on mobile platforms is challenging. To address this challenge, we introduce an efficient multi-task scene analysis approach, called EMSAFormer, that uses an RGB-D Transformer-based encoder to simultaneously perform the aforementioned tasks. Our approach builds upon the previously published EMSANet. However, we show that the dual CNN-based encoder of EMSANet can be replaced with a single Transformer-based encoder. To achieve this, we investigate how information from both RGB and depth data can be effectively incorporated in a single encoder. To accelerate inference on robotic hardware, we provide a custom NVIDIA TensorRT extension enabling highly optimization for our EMSAFormer approach. Through extensive experiments on the commonly used indoor datasets NYUv2, SUNRGB-D, and ScanNet, we show that our approach achieves state-of-the-art performance while still enabling inference with up to 39.1 FPS on an NVIDIA Jetson AGX Orin 32 GB.

Daniel Seichter, Benedict Stephan, Söhnke Benedikt Fischedick et al.

As the application scenarios of mobile robots are getting more complex and challenging, scene understanding becomes increasingly crucial. A mobile robot that is supposed to operate autonomously in indoor environments must have precise knowledge about what objects are present, where they are, what their spatial extent is, and how they can be reached; i.e., information about free space is also crucial. Panoptic mapping is a powerful instrument providing such information. However, building 3D panoptic maps with high spatial resolution is challenging on mobile robots, given their limited computing capabilities. In this paper, we propose PanopticNDT - an efficient and robust panoptic mapping approach based on occupancy normal distribution transform (NDT) mapping. We evaluate our approach on the publicly available datasets Hypersim and ScanNetV2. The results reveal that our approach can represent panoptic information at a higher level of detail than other state-of-the-art approaches while enabling real-time panoptic mapping on mobile robots. Finally, we prove the real-world applicability of PanopticNDT with qualitative results in a domestic application.

Dustin Aganian, Mona Köhler, Sebastian Baake et al.

As the use of collaborative robots (cobots) in industrial manufacturing continues to grow, human action recognition for effective human-robot collaboration becomes increasingly important. This ability is crucial for cobots to act autonomously and assist in assembly tasks. Recently, skeleton-based approaches are often used as they tend to generalize better to different people and environments. However, when processing skeletons alone, information about the objects a human interacts with is lost. Therefore, we present a novel approach of integrating object information into skeleton-based action recognition. We enhance two state-of-the-art methods by treating object centers as further skeleton joints. Our experiments on the assembly dataset IKEA ASM show that our approach improves the performance of these state-of-the-art methods to a large extent when combining skeleton joints with objects predicted by a state-of-the-art instance segmentation model. Our research sheds light on the benefits of combining skeleton joints with object information for human action recognition in assembly tasks. We analyze the effect of the object detector on the combination for action classification and discuss the important factors that must be taken into account.

Dustin Aganian, Mona Köhler, Benedict Stephan et al.

As collaborative robots (cobots) continue to gain popularity in industrial manufacturing, effective human-robot collaboration becomes crucial. Cobots should be able to recognize human actions to assist with assembly tasks and act autonomously. To achieve this, skeleton-based approaches are often used due to their ability to generalize across various people and environments. Although body skeleton approaches are widely used for action recognition, they may not be accurate enough for assembly actions where the worker's fingers and hands play a significant role. To address this limitation, we propose a method in which less detailed body skeletons are combined with highly detailed hand skeletons. We investigate CNNs and transformers, the latter of which are particularly adept at extracting and combining important information from both skeleton types using attention. This paper demonstrates the effectiveness of our proposed approach in enhancing action recognition in assembly scenarios.

Timo Sämann, Ahmed Mostafa Hammam, Andrei Bursuc et al.

Averaging predictions of a deep ensemble of networks is apopular and effective method to improve predictive performance andcalibration in various benchmarks and Kaggle competitions. However, theruntime and training cost of deep ensembles grow linearly with the size ofthe ensemble, making them unsuitable for many applications. Averagingensemble weights instead of predictions circumvents this disadvantageduring inference and is typically applied to intermediate checkpoints ofa model to reduce training cost. Albeit effective, only few works haveimproved the understanding and the performance of weight averaging.Here, we revisit this approach and show that a simple weight fusion (WF)strategy can lead to a significantly improved predictive performance andcalibration. We describe what prerequisites the weights must meet interms of weight space, functional space and loss. Furthermore, we presenta new test method (called oracle test) to measure the functional spacebetween weights. We demonstrate the versatility of our WF strategy acrossstate of the art segmentation CNNs and Transformers as well as real worlddatasets such as BDD100K and Cityscapes. We compare WF with similarapproaches and show our superiority for in- and out-of-distribution datain terms of predictive performance and calibration.

Markus Eisenbach, Jannik Lübberstedt, Dustin Aganian et al.

Person re-identification plays a key role in applications where a mobile robot needs to track its users over a long period of time, even if they are partially unobserved for some time, in order to follow them or be available on demand. In this context, deep-learning based real-time feature extraction on a mobile robot is often performed on special-purpose devices whose computational resources are shared for multiple tasks. Therefore, the inference speed has to be taken into account. In contrast, person re-identification is often improved by architectural changes that come at the cost of significantly slowing down inference. Attention blocks are one such example. We will show that some well-performing attention blocks used in the state of the art are subject to inference costs that are far too high to justify their use for mobile robotic applications. As a consequence, we propose an attention block that only slightly affects the inference speed while keeping up with much deeper networks or more complex attention blocks in terms of re-identification accuracy. We perform extensive neural architecture search to derive rules at which locations this attention block should be integrated into the architecture in order to achieve the best trade-off between speed and accuracy. Finally, we confirm that the best performing configuration on a re-identification benchmark also performs well on an indoor robotic dataset.

Timo Sämann, Horst-Michael Groß

Ensuring safety in automated driving is a major challenge for the automotive industry. Special attention is paid to artificial intelligence, in particular to Deep Neural Networks (DNNs), which is considered a key technology in the realization of highly automated driving. DNNs learn from training data, which means that they only achieve good accuracy within the underlying data distribution of the training data. When leaving the training domain, a distributional shift is caused, which can lead to a drastic reduction of accuracy. In this work, we present a proof of concept for a safety mechanism that can detect the leaving of the domain online, i.e. at runtime. In our experiments with the Synthia data set we can show that a 100 % correct detection of whether the input data is inside or outside the domain is achieved. The ability to detect when the vehicle leaves the domain can be an important requirement for certification.

Thomas Schnürer, Malte Probst, Horst-Michael Gross

Reinforcement Learning (RL) can enable agents to learn complex tasks. However, it is difficult to interpret the knowledge and reuse it across tasks. Inductive biases can address such issues by explicitly providing generic yet useful decomposition that is otherwise difficult or expensive to learn implicitly. For example, object-centered approaches decompose a high dimensional observation into individual objects. Expanding on this, we utilize an inductive bias for explicit object-centered knowledge separation that provides further decomposition into semantic representations and dynamics knowledge. For this, we introduce a semantic module that predicts an objects' semantic state based on its context. The resulting affordance-like object state can then be used to enrich perceptual object representations. With a minimal setup and an environment that enables puzzle-like tasks, we demonstrate the feasibility and benefits of this approach. Specifically, we compare three different methods of integrating semantic representations into a model-based RL architecture. Our experiments show that the degree of explicitness in knowledge separation correlates with faster learning, better accuracy, better generalization, and better interpretability.

Söhnke Benedikt Fischedick, Daniel Seichter, Benedict Stephan et al.

In domestic environments, robots require a comprehensive understanding of their surroundings to interact effectively and intuitively with untrained humans. In this paper, we propose DVEFormer - an efficient RGB-D Transformer-based approach that predicts dense text-aligned visual embeddings (DVE) via knowledge distillation. Instead of directly performing classical semantic segmentation with fixed predefined classes, our method uses teacher embeddings from Alpha-CLIP to guide our efficient student model DVEFormer in learning fine-grained pixel-wise embeddings. While this approach still enables classical semantic segmentation, e.g., via linear probing, it further enables flexible text-based querying and other applications, such as creating comprehensive 3D maps. Evaluations on common indoor datasets demonstrate that our approach achieves competitive performance while meeting real-time requirements, operating at 26.3 FPS for the full model and 77.0 FPS for a smaller variant on an NVIDIA Jetson AGX Orin. Additionally, we show qualitative results that highlight the effectiveness and possible use cases in real-world applications. Overall, our method serves as a drop-in replacement for traditional segmentation approaches while enabling flexible natural-language querying and seamless integration into 3D mapping pipelines for mobile robotics.

Mona Köhler, Markus Eisenbach, Horst-Michael Gross

Humans are able to learn to recognize new objects even from a few examples. In contrast, training deep-learning-based object detectors requires huge amounts of annotated data. To avoid the need to acquire and annotate these huge amounts of data, few-shot object detection aims to learn from few object instances of new categories in the target domain. In this survey, we provide an overview of the state of the art in few-shot object detection. We categorize approaches according to their training scheme and architectural layout. For each type of approaches, we describe the general realization as well as concepts to improve the performance on novel categories. Whenever appropriate, we give short takeaways regarding these concepts in order to highlight the best ideas. Eventually, we introduce commonly used datasets and their evaluation protocols and analyze reported benchmark results. As a result, we emphasize common challenges in evaluation and identify the most promising current trends in this emerging field of few-shot object detection.

Daniel Seichter, Mona Köhler, Benjamin Lewandowski et al.

Analyzing scenes thoroughly is crucial for mobile robots acting in different environments. Semantic segmentation can enhance various subsequent tasks, such as (semantically assisted) person perception, (semantic) free space detection, (semantic) mapping, and (semantic) navigation. In this paper, we propose an efficient and robust RGB-D segmentation approach that can be optimized to a high degree using NVIDIA TensorRT and, thus, is well suited as a common initial processing step in a complex system for scene analysis on mobile robots. We show that RGB-D segmentation is superior to processing RGB images solely and that it can still be performed in real time if the network architecture is carefully designed. We evaluate our proposed Efficient Scene Analysis Network (ESANet) on the common indoor datasets NYUv2 and SUNRGB-D and show that we reach state-of-the-art performance while enabling faster inference. Furthermore, our evaluation on the outdoor dataset Cityscapes shows that our approach is suitable for other areas of application as well. Finally, instead of presenting benchmark results only, we also show qualitative results in one of our indoor application scenarios.

Alexander Katzmann, Oliver Taubmann, Stephen Ahmad et al.

Clinical decision support using deep neural networks has become a topic of steadily growing interest. While recent work has repeatedly demonstrated that deep learning offers major advantages for medical image classification over traditional methods, clinicians are often hesitant to adopt the technology because its underlying decision-making process is considered to be intransparent and difficult to comprehend. In recent years, this has been addressed by a variety of approaches that have successfully contributed to providing deeper insight. Most notably, additive feature attribution methods are able to propagate decisions back into the input space by creating a saliency map which allows the practitioner to "see what the network sees." However, the quality of the generated maps can become poor and the images noisy if only limited data is available - a typical scenario in clinical contexts. We propose a novel decision explanation scheme based on CycleGAN activation maximization which generates high-quality visualizations of classifier decisions even in smaller data sets. We conducted a user study in which we evaluated our method on the LIDC dataset for lung lesion malignancy classification, the BreastMNIST dataset for ultrasound image breast cancer detection, as well as two subsets of the CIFAR-10 dataset for RBG image object recognition. Within this user study, our method clearly outperformed existing approaches on the medical imaging datasets and ranked second in the natural image setting. With our approach we make a significant contribution towards a better understanding of clinical decision support systems based on deep neural networks and thus aim to foster overall clinical acceptance.

Kai Fischer, Martin Simon, Florian Oelsner et al.

Robust point cloud registration in real-time is an important prerequisite for many mapping and localization algorithms. Traditional methods like ICP tend to fail without good initialization, insufficient overlap or in the presence of dynamic objects. Modern deep learning based registration approaches present much better results, but suffer from a heavy run-time. We overcome these drawbacks by introducing StickyPillars, a fast, accurate and extremely robust deep middle-end 3D feature matching method on point clouds. It uses graph neural networks and performs context aggregation on sparse 3D key-points with the aid of transformer based multi-head self and cross-attention. The network output is used as the cost for an optimal transport problem whose solution yields the final matching probabilities. The system does not rely on hand crafted feature descriptors or heuristic matching strategies. We present state-of-art art accuracy results on the registration problem demonstrated on the KITTI dataset while being four times faster then leading deep methods. Furthermore, we integrate our matching system into a LiDAR odometry pipeline yielding most accurate results on the KITTI odometry dataset. Finally, we demonstrate robustness on KITTI odometry. Our method remains stable in accuracy where state-of-the-art procedures fail on frame drops and higher speeds.

Martin Simon, Stefan Milz, Karl Amende et al.

Lidar based 3D object detection is inevitable for autonomous driving, because it directly links to environmental understanding and therefore builds the base for prediction and motion planning. The capacity of inferencing highly sparse 3D data in real-time is an ill-posed problem for lots of other application areas besides automated vehicles, e.g. augmented reality, personal robotics or industrial automation. We introduce Complex-YOLO, a state of the art real-time 3D object detection network on point clouds only. In this work, we describe a network that expands YOLOv2, a fast 2D standard object detector for RGB images, by a specific complex regression strategy to estimate multi-class 3D boxes in Cartesian space. Thus, we propose a specific Euler-Region-Proposal Network (E-RPN) to estimate the pose of the object by adding an imaginary and a real fraction to the regression network. This ends up in a closed complex space and avoids singularities, which occur by single angle estimations. The E-RPN supports to generalize well during training. Our experiments on the KITTI benchmark suite show that we outperform current leading methods for 3D object detection specifically in terms of efficiency. We achieve state of the art results for cars, pedestrians and cyclists by being more than five times faster than the fastest competitor. Further, our model is capable of estimating all eight KITTI-classes, including Vans, Trucks or sitting pedestrians simultaneously with high accuracy.