Markus Vincze

Pau Montagut Bofi, Mario García Blasco, Tessa Pulli et al.

Fine-tuning Vision-Language-Action (VLA) models for mobile manipulators with heterogeneous joint spaces can produce a counterintuitive result: the checkpoint with the lowest aggregate MSE is not the one that performs best on the real robot. We argue this is a predictable consequence of collapsing heterogeneous joint groups (arm, gripper, head, wheeled base) into a single metric, where easy-to-predict joints can mask joints that still fail. We fine-tune SmolVLA (450M, action-expert only) on the 11-DoF Toyota HSR and compare it against $π_{0.5}$ (3.3B), a stronger pretrained baseline. Per-group analysis exposes two patterns: in SmolVLA, the mobile base converges slowest and limits overall performance. In expert-only fine-tuning of $π_{0.5}$ (training only the action head, backbone frozen), total MSE drops below the baseline but arm accuracy degrades. On 60 real-robot trials (20 per model), $π_{0.5}$ 80k (4.0/4) significantly outperforms both fine-tuned variants (expert-only 3k: 3.75/4; HSR-SmolVLA: 3.5/4; Mann-Whitney $p \leq 0.010$), despite expert-only 3k having the lowest total MSE. This separation is most consistent with the offline arm-group error, not total MSE or base-group error. We conclude that per-group error is a more reliable signal than total MSE for checkpoint selection on robots with heterogeneous action spaces. Code: https://github.com/paumontagut/per-group-mse-vla

Stefan Thalhammer, Dominik Bauer, Peter Hönig et al.

Object pose estimation is a core perception task that enables, for example, object grasping and scene understanding. The widely available, inexpensive and high-resolution RGB sensors and CNNs that allow for fast inference based on this modality make monocular approaches especially well suited for robotics applications. We observe that previous surveys on object pose estimation establish the state of the art for varying modalities, single- and multi-view settings, and datasets and metrics that consider a multitude of applications. We argue, however, that those works' broad scope hinders the identification of open challenges that are specific to monocular approaches and the derivation of promising future challenges for their application in robotics. By providing a unified view on recent publications from both robotics and computer vision, we find that occlusion handling, novel pose representations, and formalizing and improving category-level pose estimation are still fundamental challenges that are highly relevant for robotics. Moreover, to further improve robotic performance, large object sets, novel objects, refractive materials, and uncertainty estimates are central, largely unsolved open challenges. In order to address them, ontological reasoning, deformability handling, scene-level reasoning, realistic datasets, and the ecological footprint of algorithms need to be improved.

Stefan Thalhammer, Timothy Patten, Markus Vincze

State-of-the-art object pose estimation handles multiple instances in a test image by using multi-model formulations: detection as a first stage and then separately trained networks per object for 2D-3D geometric correspondence prediction as a second stage. Poses are subsequently estimated using the Perspective-n-Points algorithm at runtime. Unfortunately, multi-model formulations are slow and do not scale well with the number of object instances involved. Recent approaches show that direct 6D object pose estimation is feasible when derived from the aforementioned geometric correspondences. We present an approach that learns an intermediate geometric representation of multiple objects to directly regress 6D poses of all instances in a test image. The inherent end-to-end trainability overcomes the requirement of separately processing individual object instances. By calculating the mutual Intersection-over-Unions, pose hypotheses are clustered into distinct instances, which achieves negligible runtime overhead with respect to the number of object instances. Results on multiple challenging standard datasets show that the pose estimation performance is superior to single-model state-of-the-art approaches despite being more than ~35 times faster. We additionally provide an analysis showing real-time applicability (>24 fps) for images where more than 90 object instances are present. Further results show the advantage of supervising geometric-correspondence-based object pose estimation with the 6D pose.

Xiaomin Lin, Vivek Mange, Arjun Suresh et al.

Oysters are a vital keystone species in coastal ecosystems, providing significant economic, environmental, and cultural benefits. As the importance of oysters grows, so does the relevance of autonomous systems for their detection and monitoring. However, current monitoring strategies often rely on destructive methods. While manual identification of oysters from video footage is non-destructive, it is time-consuming, requires expert input, and is further complicated by the challenges of the underwater environment. To address these challenges, we propose a novel pipeline using stable diffusion to augment a collected real dataset with realistic synthetic data. This method enhances the dataset used to train a YOLOv10-based vision model. The model is then deployed and tested on an edge platform in underwater robotics, achieving a state-of-the-art 0.657 mAP@50 for oyster detection on the Aqua2 platform.

Stefan Thalhammer, Peter Hönig, Jean-Baptiste Weibel et al.

Object pose estimation is a non-trivial task that enables robotic manipulation, bin picking, augmented reality, and scene understanding, to name a few use cases. Monocular object pose estimation gained considerable momentum with the rise of high-performing deep learning-based solutions and is particularly interesting for the community since sensors are inexpensive and inference is fast. Prior works establish the comprehensive state of the art for diverse pose estimation problems. Their broad scopes make it difficult to identify promising future directions. We narrow down the scope to the problem of single-shot monocular 6D object pose estimation, which is commonly used in robotics, and thus are able to identify such trends. By reviewing recent publications in robotics and computer vision, the state of the art is established at the union of both fields. Following that, we identify promising research directions in order to help researchers to formulate relevant research ideas and effectively advance the state of the art. Findings include that methods are sophisticated enough to overcome the domain shift and that occlusion handling is a fundamental challenge. We also highlight problems such as novel object pose estimation and challenging materials handling as central challenges to advance robotics.

Hrishikesh Gupta, Stefan Thalhammer, Markus Leitner et al.

Transparent objects are common in day-to-day life and hence find many applications that require robot grasping. Many solutions toward object grasping exist for non-transparent objects. However, due to the unique visual properties of transparent objects, standard 3D sensors produce noisy or distorted measurements. Modern approaches tackle this problem by either refining the noisy depth measurements or using some intermediate representation of the depth. Towards this, we study deep learning 6D pose estimation from RGB images only for transparent object grasping. To train and test the suitability of RGB-based object pose estimation, we construct a dataset of RGB-only images with 6D pose annotations. The experiments demonstrate the effectiveness of RGB image space for grasping transparent objects.

Konstantin Röhrl, Dominik Bauer, Timothy Patten et al.

Tracking an object's 6D pose, while either the object itself or the observing camera is moving, is important for many robotics and augmented reality applications. While exploiting temporal priors eases this problem, object-specific knowledge is required to recover when tracking is lost. Under the tight time constraints of the tracking task, RGB(D)-based methods are often conceptionally complex or rely on heuristic motion models. In comparison, we propose to simplify object tracking to a reinforced point cloud (depth only) alignment task. This allows us to train a streamlined approach from scratch with limited amounts of sparse 3D point clouds, compared to the large datasets of diverse RGBD sequences required in previous works. We incorporate temporal frame-to-frame registration with object-based recovery by frame-to-model refinement using a reinforcement learning (RL) agent that jointly solves for both objectives. We also show that the RL agent's uncertainty and a rendering-based mask propagation are effective reinitialization triggers.

Philipp Ausserlechner, David Haberger, Stefan Thalhammer et al.

As robotic systems increasingly encounter complex and unconstrained real-world scenarios, there is a demand to recognize diverse objects. The state-of-the-art 6D object pose estimation methods rely on object-specific training and therefore do not generalize to unseen objects. Recent novel object pose estimation methods are solving this issue using task-specific fine-tuned CNNs for deep template matching. This adaptation for pose estimation still requires expensive data rendering and training procedures. MegaPose for example is trained on a dataset consisting of two million images showing 20,000 different objects to reach such generalization capabilities. To overcome this shortcoming we introduce ZS6D, for zero-shot novel object 6D pose estimation. Visual descriptors, extracted using pre-trained Vision Transformers (ViT), are used for matching rendered templates against query images of objects and for establishing local correspondences. These local correspondences enable deriving geometric correspondences and are used for estimating the object's 6D pose with RANSAC-based PnP. This approach showcases that the image descriptors extracted by pre-trained ViTs are well-suited to achieve a notable improvement over two state-of-the-art novel object 6D pose estimation methods, without the need for task-specific fine-tuning. Experiments are performed on LMO, YCBV, and TLESS. In comparison to one of the two methods we improve the Average Recall on all three datasets and compared to the second method we improve on two datasets.

Gerald Ebmer, Adam Loch, Minh Nhat Vu et al.

Real-time applications for autonomous operations depend largely on fast and robust vision-based localization systems. Since image processing tasks require processing large amounts of data, the computational resources often limit the performance of other processes. To overcome this limitation, traditional marker-based localization systems are widely used since they are easy to integrate and achieve reliable accuracy. However, classical marker-based localization systems significantly depend on standard cameras with low frame rates, which often lack accuracy due to motion blur. In contrast, event-based cameras provide high temporal resolution and a high dynamic range, which can be utilized for fast localization tasks, even under challenging visual conditions. This paper proposes a simple but effective event-based pose estimation system using active LED markers (ALM) for fast and accurate pose estimation. The proposed algorithm is able to operate in real time with a latency below \SI{0.5}{\milli\second} while maintaining output rates of \SI{3}{\kilo \hertz}. Experimental results in static and dynamic scenarios are presented to demonstrate the performance of the proposed approach in terms of computational speed and absolute accuracy, using the OptiTrack system as the basis for measurement.

Marc Glocker, Peter Hönig, Matthias Hirschmanner et al.

We present an embodied robotic system with an LLM-driven agent-orchestration architecture for autonomous household object management. The system integrates memory-augmented task planning, enabling robots to execute high-level user commands while tracking past actions. It employs three specialized agents: a routing agent, a task planning agent, and a knowledge base agent, each powered by task-specific LLMs. By leveraging in-context learning, our system avoids the need for explicit model training. RAG enables the system to retrieve context from past interactions, enhancing long-term object tracking. A combination of Grounded SAM and LLaMa3.2-Vision provides robust object detection, facilitating semantic scene understanding for task planning. Evaluation across three household scenarios demonstrates high task planning accuracy and an improvement in memory recall due to RAG. Specifically, Qwen2.5 yields best performance for specialized agents, while LLaMA3.1 excels in routing tasks. The source code is available at: https://github.com/marc1198/chat-hsr.

Tessa Pulli, Jean-Baptiste Weibel, Peter Hönig et al.

6D object pose estimation plays a crucial role in scene understanding for applications such as robotics and augmented reality. To support the needs of ever-changing object sets in such context, modern zero-shot object pose estimators were developed to not require object-specific training but only rely on CAD models. Such models are hard to obtain once deployed, and a continuously changing and growing set of objects makes it harder to reliably identify the instance model of interest. To address this challenge, we introduce an Open-Set CAD Retrieval from a Language Prompt and a Single Image (OSCAR), a novel training-free method that retrieves a matching object model from an unlabeled 3D object database. During onboarding, OSCAR generates multi-view renderings of database models and annotates them with descriptive captions using an image captioning model. At inference, GroundedSAM detects the queried object in the input image, and multi-modal embeddings are computed for both the Region-of-Interest and the database captions. OSCAR employs a two-stage retrieval: text-based filtering using CLIP identifies candidate models, followed by image-based refinement using DINOv2 to select the most visually similar object. In our experiments we demonstrate that OSCAR outperforms all state-of-the-art methods on the cross-domain 3D model retrieval benchmark MI3DOR. Furthermore, we demonstrate OSCAR's direct applicability in automating object model sourcing for 6D object pose estimation. We propose using the most similar object model for pose estimation if the exact instance is not available and show that OSCAR achieves an average precision of 90.48\% during object retrieval on the YCB-V object dataset. Moreover, we demonstrate that the most similar object model can be utilized for pose estimation using Megapose achieving better results than a reconstruction-based approach.

Tessa Pulli, Stefan Thalhammer, Simon Schwaiger et al.

Robots are increasingly envisioned to interact in real-world scenarios, where they must continuously adapt to new situations. To detect and grasp novel objects, zero-shot pose estimators determine poses without prior knowledge. Recently, vision language models (VLMs) have shown considerable advances in robotics applications by establishing an understanding between language input and image input. In our work, we take advantage of VLMs zero-shot capabilities and translate this ability to 6D object pose estimation. We propose a novel framework for promptable zero-shot 6D object pose estimation using language embeddings. The idea is to derive a coarse location of an object based on the relevancy map of a language-embedded NeRF reconstruction and to compute the pose estimate with a point cloud registration method. Additionally, we provide an analysis of LERF's suitability for open-set object pose estimation. We examine hyperparameters, such as activation thresholds for relevancy maps and investigate the zero-shot capabilities on an instance- and category-level. Furthermore, we plan to conduct robotic grasping experiments in a real-world setting.

Hrishikesh Gupta, Stefan Thalhammer, Jean-Baptiste Weibel et al.

Transparent objects are ubiquitous in daily life, making their perception and robotics manipulation important. However, they present a major challenge due to their distinct refractive and reflective properties when it comes to accurately estimating the 6D pose. To solve this, we present ReFlow6D, a novel method for transparent object 6D pose estimation that harnesses the refractive-intermediate representation. Unlike conventional approaches, our method leverages a feature space impervious to changes in RGB image space and independent of depth information. Drawing inspiration from image matting, we model the deformation of the light path through transparent objects, yielding a unique object-specific intermediate representation guided by light refraction that is independent of the environment in which objects are observed. By integrating these intermediate features into the pose estimation network, we show that ReFlow6D achieves precise 6D pose estimation of transparent objects, using only RGB images as input. Our method further introduces a novel transparent object compositing loss, fostering the generation of superior refractive-intermediate features. Empirical evaluations show that our approach significantly outperforms state-of-the-art methods on TOD and Trans32K-6D datasets. Robot grasping experiments further demonstrate that ReFlow6D's pose estimation accuracy effectively translates to real-world robotics task. The source code is available at: https://github.com/StoicGilgamesh/ReFlow6D and https://github.com/StoicGilgamesh/matting_rendering.

Bernd Von Gimborn, Philipp Ausserlechner, Markus Vincze et al.

Zero-shot object pose estimation enables the retrieval of object poses from images without necessitating object-specific training. In recent approaches this is facilitated by vision foundation models (VFM), which are pre-trained models that are effectively general-purpose feature extractors. The characteristics exhibited by these VFMs vary depending on the training data, network architecture, and training paradigm. The prevailing choice in this field are self-supervised Vision Transformers (ViT). This study assesses the influence of Latent Diffusion Model (LDM) backbones on zero-shot pose estimation. In order to facilitate a comparison between the two families of models on a common ground we adopt and modify a recent approach. Therefore, a template-based multi-staged method for estimating poses in a zero-shot fashion using LDMs is presented. The efficacy of the proposed approach is empirically evaluated on three standard datasets for object-specific 6DoF pose estimation. The experiments demonstrate an Average Recall improvement of up to 27% over the ViT baseline. The source code is available at: https://github.com/BvG1993/DZOP.

Peter Hönig, Stefan Thalhammer, Jean-Baptiste Weibel et al.

Object manipulation requires accurate object pose estimation. In open environments, robots encounter unknown objects, which requires semantic understanding in order to generalize both to known categories and beyond. To resolve this challenge, we present SCOPE, a diffusion-based category-level object pose estimation model that eliminates the need for discrete category labels by leveraging DINOv2 features as continuous semantic priors. By combining these DINOv2 features with photorealistic training data and a noise model for point normals, we reduce the Sim2Real gap in category-level object pose estimation. Furthermore, injecting the continuous semantic priors via cross-attention enables SCOPE to learn canonicalized object coordinate systems across object instances beyond the distribution of known categories. SCOPE outperforms the current state of the art in synthetically trained category-level object pose estimation, achieving a relative improvement of 31.9\% on the 5$^\circ$5cm metric. Additional experiments on two instance-level datasets demonstrate generalization beyond known object categories, enabling grasping of unseen objects from unknown categories with a success rate of up to 100\%. Code available: https://github.com/hoenigpeter/scope.

Pau Amargant, Peter Hönig, Markus Vincze

The verification of successful grasps is a crucial aspect of robot manipulation, particularly when handling deformable objects. Traditional methods relying on force and tactile sensors often struggle with deformable and non-rigid objects. In this work, we present a vision-based approach for grasp verification to determine whether the robotic gripper has successfully grasped an object. Our method employs a two-stage architecture; first YOLO-based object detection model to detect and locate the robot's gripper and then a ResNet-based classifier determines the presence of an object. To address the limitations of real-world data capture, we introduce HSR-GraspSynth, a synthetic dataset designed to simulate diverse grasping scenarios. Furthermore, we explore the use of Visual Question Answering capabilities as a zero-shot baseline to which we compare our model. Experimental results demonstrate that our approach achieves high accuracy in real-world environments, with potential for integration into grasping pipelines. Code and datasets are publicly available at https://github.com/pauamargant/HSR-GraspSynth .

Peter Hönig, Stefan Thalhammer, Jean-Baptiste Weibel et al.

Recent advances in machine learning have greatly benefited object detection and 6D pose estimation. However, textureless and metallic objects still pose a significant challenge due to few visual cues and the texture bias of CNNs. To address his issue, we propose a strategy for inducing a shape bias to CNN training. In particular, by randomizing textures applied to object surfaces during data rendering, we create training data without consistent textural cues. This methodology allows for seamless integration into existing data rendering engines, and results in negligible computational overhead for data rendering and network training. Our findings demonstrate that the shape bias we induce via randomized texturing, improves over existing approaches using style transfer. We evaluate with three detectors and two pose estimators. For the most recent object detector and for pose estimation in general, estimation accuracy improves for textureless and metallic objects. Additionally we show that our approach increases the pose estimation accuracy in the presence of image noise and strong illumination changes. Code and datasets are publicly available at github.com/hoenigpeter/randomized_texturing.

Dominik Bauer, Timothy Patten, Markus Vincze

Accurate and robust object pose estimation for robotics applications requires verification and refinement steps. In this work, we propose to integrate hypotheses verification with object pose refinement guided by physics simulation. This allows the physical plausibility of individual object pose estimates and the stability of the estimated scene to be considered in a unified optimization. The proposed method is able to adapt to scenes of multiple objects and efficiently focuses on refining the most promising object poses in multi-hypotheses scenarios. We call this integrated approach VeREFINE and evaluate it on three datasets with varying scene complexity. The generality of the approach is shown by using three state-of-the-art pose estimators and three baseline refiners. Results show improvements over all baselines and on all datasets. Furthermore, our approach is applied in real-world grasping experiments and outperforms competing methods in terms of grasp success rate. Code is publicly available at github.com/dornik/verefine.

Peter Hönig, Stefan Thalhammer, Markus Vincze

Estimating 2D-3D correspondences between RGB images and 3D space is a fundamental problem in 6D object pose estimation. Recent pose estimators use dense correspondence maps and Point-to-Point algorithms to estimate object poses. The accuracy of pose estimation depends heavily on the quality of the dense correspondence maps and their ability to withstand occlusion, clutter, and challenging material properties. Currently, dense correspondence maps are estimated using image-to-image translation models based on GANs, Autoencoders, or direct regression models. However, recent advancements in image-to-image translation have led to diffusion models being the superior choice when evaluated on benchmarking datasets. In this study, we compare image-to-image translation networks based on GANs and diffusion models for the downstream task of 6D object pose estimation. Our results demonstrate that the diffusion-based image-to-image translation model outperforms the GAN, revealing potential for further improvements in 6D object pose estimation models.

Peter Hönig, Matthias Hirschmanner, Markus Vincze

Object pose estimation enables a variety of tasks in computer vision and robotics, including scene understanding and robotic grasping. The complexity of a pose estimation task depends on the unknown variables related to the target object. While instance-level methods already excel for opaque and Lambertian objects, category-level and open-set methods, where texture, shape, and size are partially or entirely unknown, still struggle with these basic material properties. Since texture is unknown in these scenarios, it cannot be used for disambiguating object symmetries, another core challenge of 6D object pose estimation. The complexity of estimating 6D poses with such a manifold of unknowns led to various datasets, accuracy metrics, and algorithmic solutions. This paper compares datasets, accuracy metrics, and algorithms for solving 6D pose estimation on the category-level. Based on this comparison, we analyze how to bridge category-level and open-set object pose estimation to reach generalization and provide actionable recommendations.

Melvin Reka, Tessa Pulli, Markus Vincze

6D object pose estimation for unseen objects is essential in robotics but traditionally relies on trained models that require large datasets, high computational costs, and struggle to generalize. Zero-shot approaches eliminate the need for training but depend on pre-existing 3D object models, which are often impractical to obtain. To address this, we propose a language-guided few-shot 3D reconstruction method, reconstructing a 3D mesh from few input images. In the proposed pipeline, receives a set of input images and a language query. A combination of GroundingDINO and Segment Anything Model outputs segmented masks from which a sparse point cloud is reconstructed with VGGSfM. Subsequently, the mesh is reconstructed with the Gaussian Splatting method SuGAR. In a final cleaning step, artifacts are removed, resulting in the final 3D mesh of the queried object. We evaluate the method in terms of accuracy and quality of the geometry and texture. Furthermore, we study the impact of imaging conditions such as viewing angle, number of input images, and image overlap on 3D object reconstruction quality, efficiency, and computational scalability.

Tessa Pulli, Peter Hönig, Stefan Thalhammer et al.

Object pose estimation of transparent objects remains a challenging task in the field of robot vision due to the immense influence of lighting, background, and reflections. However, the edges of clear objects have the highest contrast, which leads to stable and prominent features. We propose a novel approach by incorporating edge detection in a pre-processing step for the tasks of object detection and object pose estimation. We conducted experiments to investigate the effect of edge detectors on transparent objects. We examine the performance of the state-of-the-art 6D object pose estimation pipeline GDR-Net and the object detector YOLOX when applying different edge detectors as pre-processing steps (i.e., Canny edge detection with and without color information, and holistically-nested edges (HED)). We evaluate the physically-based rendered dataset Trans6D-32 K of transparent objects with parameters proposed by the BOP Challenge. Our results indicate that applying edge detection as a pre-processing enhances performance for certain objects.

Stefan Thalhammer, Jean-Baptiste Weibel, Markus Vincze et al.

Object pose estimation is important for object manipulation and scene understanding. In order to improve the general applicability of pose estimators, recent research focuses on providing estimates for novel objects, that is objects unseen during training. Such works use deep template matching strategies to retrieve the closest template connected to a query image. This template retrieval implicitly provides object class and pose. Despite the recent success and improvements of Vision Transformers over CNNs for many vision tasks, the state of the art uses CNN-based approaches for novel object pose estimation. This work evaluates and demonstrates the differences between self-supervised CNNs and Vision Transformers for deep template matching. In detail, both types of approaches are trained using contrastive learning to match training images against rendered templates of isolated objects. At test time, such templates are matched against query images of known and novel objects under challenging settings, such as clutter, occlusion and object symmetries, using masked cosine similarity. The presented results not only demonstrate that Vision Transformers improve in matching accuracy over CNNs, but also that for some cases pre-trained Vision Transformers do not need fine-tuning to do so. Furthermore, we highlight the differences in optimization and network architecture when comparing these two types of network for deep template matching.

Dominik Bauer, Timothy Patten, Markus Vincze

Observational noise, inaccurate segmentation and ambiguity due to symmetry and occlusion lead to inaccurate object pose estimates. While depth- and RGB-based pose refinement approaches increase the accuracy of the resulting pose estimates, they are susceptible to ambiguity in the observation as they consider visual alignment. We propose to leverage the fact that we often observe static, rigid scenes. Thus, the objects therein need to be under physically plausible poses. We show that considering plausibility reduces ambiguity and, in consequence, allows poses to be more accurately predicted in cluttered environments. To this end, we extend a recent RL-based registration approach towards iterative refinement of object poses. Experiments on the LINEMOD and YCB-VIDEO datasets demonstrate the state-of-the-art performance of our depth-based refinement approach.

Adam Loch, Germain Haessig, Markus Vincze

Motion and dynamic environments, especially under challenging lighting conditions, are still an open issue for robust robotic applications. In this paper, we propose an end-to-end pipeline for real-time, low latency, 6 degrees-of-freedom pose estimation of fiducial markers. Instead of achieving a pose estimation through a conventional frame-based approach, we employ the high-speed abilities of event-based sensors to directly refine the spatial transformation, using consecutive events. Furthermore, we introduce a novel two-way verification process for detecting tracking errors by backtracking the estimated pose, allowing us to evaluate the quality of our tracking. This approach allows us to achieve pose estimation at a rate up to 156~kHz, while only relying on CPU resources. The average end-to-end latency of our method is 3~ms. Experimental results demonstrate outstanding potential for robotic tasks, such as visual servoing in fast action-perception loops.

Pablo Martinez-Gonzalez, Sergiu Oprea, John Alejandro Castro-Vargas et al.

Synthetic data generation has become essential in last years for feeding data-driven algorithms, which surpassed traditional techniques performance in almost every computer vision problem. Gathering and labelling the amount of data needed for these data-hungry models in the real world may become unfeasible and error-prone, while synthetic data give us the possibility of generating huge amounts of data with pixel-perfect annotations. However, most synthetic datasets lack from enough realism in their rendered images. In that context UnrealROX generation tool was presented in 2019, allowing to generate highly realistic data, at high resolutions and framerates, with an efficient pipeline based on Unreal Engine, a cutting-edge videogame engine. UnrealROX enabled robotic vision researchers to generate realistic and visually plausible data with full ground truth for a wide variety of problems such as class and instance semantic segmentation, object detection, depth estimation, visual grasping, and navigation. Nevertheless, its workflow was very tied to generate image sequences from a robotic on-board camera, making hard to generate data for other purposes. In this work, we present UnrealROX+, an improved version of UnrealROX where its decoupled and easy-to-use data acquisition system allows to quickly design and generate data in a much more flexible and customizable way. Moreover, it is packaged as an Unreal plug-in, which makes it more comfortable to use with already existing Unreal projects, and it also includes new features such as generating albedo or a Python API for interacting with the virtual environment from Deep Learning frameworks.

Michael Koller, Timothy Patten, Markus Vincze

Assistive multi-armed bandit problems can be used to model team situations between a human and an autonomous system like a domestic service robot. To account for human biases such as the risk-aversion described in the Cumulative Prospect Theory, the setting is expanded to using observable rewards. When robots leverage knowledge about the risk-averse human model they eliminate the bias and make more rational choices. We present an algorithm that increases the utility value of such human-robot teams. A brief evaluation indicates that arbitrary reward functions can be handled.

Dominik Bauer, Timothy Patten, Markus Vincze

Point cloud registration is a common step in many 3D computer vision tasks such as object pose estimation, where a 3D model is aligned to an observation. Classical registration methods generalize well to novel domains but fail when given a noisy observation or a bad initialization. Learning-based methods, in contrast, are more robust but lack in generalization capacity. We propose to consider iterative point cloud registration as a reinforcement learning task and, to this end, present a novel registration agent (ReAgent). We employ imitation learning to initialize its discrete registration policy based on a steady expert policy. Integration with policy optimization, based on our proposed alignment reward, further improves the agent's registration performance. We compare our approach to classical and learning-based registration methods on both ModelNet40 (synthetic) and ScanObjectNN (real data) and show that our ReAgent achieves state-of-the-art accuracy. The lightweight architecture of the agent, moreover, enables reduced inference time as compared to related approaches. In addition, we apply our method to the object pose estimation task on real data (LINEMOD), outperforming state-of-the-art pose refinement approaches.

Jean-Baptiste Weibel, Timothy Patten, Markus Vincze

While object semantic understanding is essential for most service robotic tasks, 3D object classification is still an open problem. Learning from artificial 3D models alleviates the cost of annotation necessary to approach this problem, but most methods still struggle with the differences existing between artificial and real 3D data. We conjecture that the cause of those issue is the fact that many methods learn directly from point coordinates, instead of the shape, as the former is hard to center and to scale under variable occlusions reliably. We introduce spherical kernel point convolutions that directly exploit the object surface, represented as a graph, and a voting scheme to limit the impact of poor segmentation on the classification results. Our proposed approach improves upon state-of-the-art methods by up to 36% when transferring from artificial objects to real objects.

Stefan Thalhammer, Markus Leitner, Timothy Patten et al.

Object pose estimation enables robots to understand and interact with their environments. Training with synthetic data is necessary in order to adapt to novel situations. Unfortunately, pose estimation under domain shift, i.e., training on synthetic data and testing in the real world, is challenging. Deep learning-based approaches currently perform best when using encoder-decoder networks but typically do not generalize to new scenarios with different scene characteristics. We argue that patch-based approaches, instead of encoder-decoder networks, are more suited for synthetic-to-real transfer because local to global object information is better represented. To that end, we present a novel approach based on a specialized feature pyramid network to compute multi-scale features for creating pose hypotheses on different feature map resolutions in parallel. Our single-shot pose estimation approach is evaluated on multiple standard datasets and outperforms the state of the art by up to 35%. We also perform grasping experiments in the real world to demonstrate the advantage of using synthetic data to generalize to novel environments.

Kiru Park, Timothy Patten, Markus Vincze

Recent methods for 6D pose estimation of objects assume either textured 3D models or real images that cover the entire range of target poses. However, it is difficult to obtain textured 3D models and annotate the poses of objects in real scenarios. This paper proposes a method, Neural Object Learning (NOL), that creates synthetic images of objects in arbitrary poses by combining only a few observations from cluttered images. A novel refinement step is proposed to align inaccurate poses of objects in source images, which results in better quality images. Evaluations performed on two public datasets show that the rendered images created by NOL lead to state-of-the-art performance in comparison to methods that use 13 times the number of real images. Evaluations on our new dataset show multiple objects can be trained and recognized simultaneously using a sequence of a fixed scene.

Mohammad Reza Loghmani, Luca Robbiano, Mirco Planamente et al.

Unsupervised Domain Adaptation (DA) exploits the supervision of a label-rich source dataset to make predictions on an unlabeled target dataset by aligning the two data distributions. In robotics, DA is used to take advantage of automatically generated synthetic data, that come with "free" annotation, to make effective predictions on real data. However, existing DA methods are not designed to cope with the multi-modal nature of RGB-D data, which are widely used in robotic vision. We propose a novel RGB-D DA method that reduces the synthetic-to-real domain shift by exploiting the inter-modal relation between the RGB and depth image. Our method consists of training a convolutional neural network to solve, in addition to the main recognition task, the pretext task of predicting the relative rotation between the RGB and depth image. To evaluate our method and encourage further research in this area, we define two benchmark datasets for object categorization and instance recognition. With extensive experiments, we show the benefits of leveraging the inter-modal relations for RGB-D DA.

Zhi Yan, Simon Schreiberhuber, Georg Halmetschlager et al.

This paper presents the perception system of a new professional cleaning robot for large public places. The proposed system is based on multiple sensors including 3D and 2D lidar, two RGB-D cameras and a stereo camera. The two lidars together with an RGB-D camera are used for dynamic object (human) detection and tracking, while the second RGB-D and stereo camera are used for detection of static objects (dirt and ground objects). A learning and reasoning module for spatial-temporal representation of the environment based on the perception pipeline is also introduced. Furthermore, a new dataset collected with the robot in several public places, including a supermarket, a warehouse and an airport, is released. Baseline results on this dataset for further research and comparison are provided. The proposed system has been fully implemented into the Robot Operating System (ROS) with high modularity, also publicly available to the community.

Timothy Patten, Kiru Park, Markus Vincze

This article presents a method for grasping novel objects by learning from experience. Successful attempts are remembered and then used to guide future grasps such that more reliable grasping is achieved over time. To generalise the learned experience to unseen objects, we introduce the dense geometric correspondence matching network (DGCM-Net). This applies metric learning to encode objects with similar geometry nearby in feature space. Retrieving relevant experience for an unseen object is thus a nearest neighbour search with the encoded feature maps. DGCM-Net also reconstructs 3D-3D correspondences using the view-dependent normalised object coordinate space to transform grasp configurations from retrieved samples to unseen objects. In comparison to baseline methods, our approach achieves an equivalent grasp success rate. However, the baselines are significantly improved when fusing the knowledge from experience with their grasp proposal strategy. Offline experiments with a grasping dataset highlight the capability to generalise within and between object classes as well as to improve success rate over time from increasing experience. Lastly, by learning task-relevant grasps, our approach can prioritise grasps that enable the functional use of objects.

Luis A. Mateos, Markus Vincze

The robot's objective is to rehabilitate the pipe joints of fresh water supply systems by crawling into water canals and applying a restoration material to repair the pipes. The robot's structure consists of six wheeled-legs, three on the front separated 120° and three on the back in the same configuration, supporting the structure along the centre of the pipe. In this configuration the robot is able to clean and seal with a rotating tool, similar to a cylindrical robot, covering the entire 3D in-pipe space.

Jean-Baptiste Weibel, Timothy Patten, Markus Vincze

Object classification with 3D data is an essential component of any scene understanding method. It has gained significant interest in a variety of communities, most notably in robotics and computer graphics. While the advent of deep learning has progressed the field of 3D object classification, most work using this data type are solely evaluated on CAD model datasets. Consequently, current work does not address the discrepancies existing between real and artificial data. In this work, we examine this gap in a robotic context by specifically addressing the problem of classification when transferring from artificial CAD models to real reconstructed objects. This is performed by training on ModelNet (CAD models) and evaluating on ScanNet (reconstructed objects). We show that standard methods do not perform well in this task. We thus introduce a method that carefully samples object parts that are reproducible under various transformations and hence robust. Using graph convolution to classify the composed graph of parts, our method significantly improves upon the baseline.

Kiru Park, Timothy Patten, Markus Vincze

Estimating the 6D pose of objects using only RGB images remains challenging because of problems such as occlusion and symmetries. It is also difficult to construct 3D models with precise texture without expert knowledge or specialized scanning devices. To address these problems, we propose a novel pose estimation method, Pix2Pose, that predicts the 3D coordinates of each object pixel without textured models. An auto-encoder architecture is designed to estimate the 3D coordinates and expected errors per pixel. These pixel-wise predictions are then used in multiple stages to form 2D-3D correspondences to directly compute poses with the PnP algorithm with RANSAC iterations. Our method is robust to occlusion by leveraging recent achievements in generative adversarial training to precisely recover occluded parts. Furthermore, a novel loss function, the transformer loss, is proposed to handle symmetric objects by guiding predictions to the closest symmetric pose. Evaluations on three different benchmark datasets containing symmetric and occluded objects show our method outperforms the state of the art using only RGB images.

Markus Suchi, Timothy Patten, David Fischinger et al.

Developing robot perception systems for recognizing objects in the real-world requires computer vision algorithms to be carefully scrutinized with respect to the expected operating domain. This demands large quantities of ground truth data to rigorously evaluate the performance of algorithms. This paper presents the EasyLabel tool for easily acquiring high quality ground truth annotation of objects at the pixel-level in densely cluttered scenes. In a semi-automatic process, complex scenes are incrementally built and EasyLabel exploits depth change to extract precise object masks at each step. We use this tool to generate the Object Cluttered Indoor Dataset (OCID) that captures diverse settings of objects, background, context, sensor to scene distance, viewpoint angle and lighting conditions. OCID is used to perform a systematic comparison of existing object segmentation methods. The baseline comparison supports the need for pixel- and object-wise annotation to progress robot vision towards realistic applications. This insight reveals the usefulness of EasyLabel and OCID to better understand the challenges that robots face in the real-world. Copyright 20XX IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

Mohammad Reza Loghmani, Mirco Planamente, Barbara Caputo et al.

Providing machines with the ability to recognize objects like humans has always been one of the primary goals of machine vision. The introduction of RGB-D cameras has paved the way for a significant leap forward in this direction thanks to the rich information provided by these sensors. However, the machine vision community still lacks an effective method to synergically use the RGB and depth data to improve object recognition. In order to take a step in this direction, we introduce a novel end-to-end architecture for RGB-D object recognition called recurrent convolutional fusion (RCFusion). Our method generates compact and highly discriminative multi-modal features by combining complementary RGB and depth information representing different levels of abstraction. Extensive experiments on two popular datasets, RGB-D Object Dataset and JHUIT-50, show that RCFusion significantly outperforms state-of-the-art approaches in both the object categorization and instance recognition tasks.

Sergey V. Alexandrov, Johann Prankl, Michael Zillich et al.

The research in dense online 3D mapping is mostly focused on the geometrical accuracy and spatial extent of the reconstructions. Their color appearance is often neglected, leading to inconsistent colors and noticeable artifacts. We rectify this by extending a state-of-the-art SLAM system to accumulate colors in HDR space. We replace the simplistic pixel intensity averaging scheme with HDR color fusion rules tailored to the incremental nature of SLAM and a noise model suitable for off-the-shelf RGB-D cameras. Our main contribution is a map-aware exposure time controller. It makes decisions based on the global state of the map and predicted camera motion, attempting to maximize the information gain of each observation. We report a set of experiments demonstrating the improved texture quality and advantages of using the custom controller that is tightly integrated in the mapping loop.

Mohammad Reza Loghmani, Barbara Caputo, Markus Vincze

The ability to recognize objects is an essential skill for a robotic system acting in human-populated environments. Despite decades of effort from the robotic and vision research communities, robots are still missing good visual perceptual systems, preventing the use of autonomous agents for real-world applications. The progress is slowed down by the lack of a testbed able to accurately represent the world perceived by the robot in-the-wild. In order to fill this gap, we introduce a large-scale, multi-view object dataset collected with an RGB-D camera mounted on a mobile robot. The dataset embeds the challenges faced by a robot in a real-life application and provides a useful tool for validating object recognition algorithms. Besides describing the characteristics of the dataset, the paper evaluates the performance of a collection of well-established deep convolutional networks on the new dataset and analyzes the transferability of deep representations from Web images to robotic data. Despite the promising results obtained with such representations, the experiments demonstrate that object classification with real-life robotic data is far from being solved. Finally, we provide a comparative study to analyze and highlight the open challenges in robot vision, explaining the discrepancies in the performance.

Markus Bajones, Astrid Weiss, Markus Vincze

Service robots for the domestic environment are intended to autonomously provide support for their users. However, state-of-the-art robots still often get stuck in failure situations leading to breakdowns in the interaction flow from which the robot cannot recover alone. We performed a multi-user Wizard-of-Oz experiment in which we manipulated the robot's behavior in such a way that it appeared unexpected and malfunctioning, and asked participants to help the robot in order to restore the interaction flow. We examined how participants reacted to the robot's error, its subsequent request for help and how it changed their perception of the robot with respect to perceived intelligence, likability, and task contribution. As interaction scenario we used a game of building Lego models performed by user dyads. In total 38 participants interacted with the robot and helped in malfunctioning situations. We report two major findings: (1) in user dyads, the user who gave the last command followed by the user who is closer is more likely to help (2) malfunctions that can be actively fixed by the user seem not to negatively impact perceived intelligence and likability ratings. This work offers insights in how far user support can be a strategy for domestic service robots to recover from repeating malfunctions.

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.

Markus Bajones, Daniel Wolf, Johann Prankl et al.

This paper presents the behaviour control of a service robot for intelligent object search in a domestic environment. A major challenge in service robotics is to enable fetch-and-carry missions that are satisfying for the user in terms of efficiency and human-oriented perception. The proposed behaviour controller provides an informed intelligent search based on a semantic segmentation framework for indoor scenes and integrates it with object recognition and grasping. Instead of manually annotating search positions in the environment, the framework automatically suggests likely locations to search for an object based on contextual information, e.g. next to tables and shelves. In a preliminary set of experiments we demonstrate that this behaviour control is as efficient as using manually annotated locations. Moreover, we argue that our approach will reduce the intensity of labour associated with programming fetch-and-carry tasks for service robots and that it will be perceived as more human-oriented.

Andreas Grünauer, Markus Vincze

In this work we show that the classification performance of high-dimensional structural MRI data with only a small set of training examples is improved by the usage of dimension reduction methods. We assessed two different dimension reduction variants: feature selection by ANOVA F-test and feature transformation by PCA. On the reduced datasets, we applied common learning algorithms using 5-fold cross-validation. Training, tuning of the hyperparameters, as well as the performance evaluation of the classifiers was conducted using two different performance measures: Accuracy, and Receiver Operating Characteristic curve (AUC). Our hypothesis is supported by experimental results.

Aitor Aldoma, Johann Prankl, Alexander Svejda et al.

This work presents a flexible system to reconstruct 3D models of objects captured with an RGB-D sensor. A major advantage of the method is that our reconstruction pipeline allows the user to acquire a full 3D model of the object. This is achieved by acquiring several partial 3D models in different sessions that are automatically merged together to reconstruct a full model. In addition, the 3D models acquired by our system can be directly used by state-of-the-art object instance recognition and object tracking modules, providing object-perception capabilities for different applications, such as human-object interaction analysis or robot grasping. The system does not impose constraints in the appearance of objects (textured, untextured) nor in the modelling setup (moving camera with static object or a turn-table setup). The proposed reconstruction system has been used to model a large number of objects resulting in metrically accurate and visually appealing 3D models.

Daniel Wolf, Markus Bajones, Johann Prankl et al.

In this paper, we propose an efficient semantic segmentation framework for indoor scenes, tailored to the application on a mobile robot. Semantic segmentation can help robots to gain a reasonable understanding of their environment, but to reach this goal, the algorithms not only need to be accurate, but also fast and robust. Therefore, we developed an optimized 3D point cloud processing framework based on a Randomized Decision Forest, achieving competitive results at sufficiently high frame rates. We evaluate the capabilities of our method on the popular NYU depth dataset and our own data and demonstrate its feasibility by deploying it on a mobile service robot, for which we could optimize an object search procedure using our results.

Markus Bader, Johann Prankl, Markus Vincze

Humanoid robots without internal sensors such as a compass tend to lose their orientation after a fall. Furthermore, re-initialisation is often ambiguous due to symmetric man-made environments. The room-awareness module proposed here is inspired by the results of psychological experiments and improves existing self-localization strategies by mapping and matching the visual background with colour histograms. The matching algorithm uses a particle-filter to generate hypotheses of the viewing directions independent of the self-localization algorithm and generates confidence values for various possible poses. The robot's behaviour controller uses those confidence values to control self-localization algorithm to converge to the most likely pose and prevents the algorithm from getting stuck in local minima. Experiments with a symmetric Standard Platform League RoboCup playing field with a simulated and a real humanoid NAO robot show the significant improvement of the system.