The Ninth Robotic Grasping and Manipulation Competition (RGMC) took place in Yokohama, Japan, during the 2024 IEEE/RAS International Conference on Robotics and Automation (ICRA). The series of RGMC events started in 2016 at the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) with strong support from the conference’s organization committee, and since then they have been held each year at ICRA or IROS. Across the editions, RGMC engaged the community in solving the open challenges associated with various robotic grasping and manipulation tasks for manufacturing, service robots, and logistics, and in advancing research and technology towards more realistic scenarios that can be encountered in daily activities at home or in warehouses. These tasks include assembling and disassembling boards, hand-in-hand grasping, picking and placing various objects, pouring liquids into a cup, bin picking, rearranging and setting formal tables, folding and unfolding cloths, and receiving objects handed over by a person. The goal of RGMC across these tasks is to assess the autonomous manipulation capabilities of a robotic arm when dealing with unknown or novel objects with varying physical properties and when handling scenarios with various degrees of uncertainty caused by a cluttered scene, random initial configurations, or human behaviors when interacting with the robot. For example, objects can vary in their shapes, appearances, transparency, deformability, and weight. RGMC challenges participants to design solutions that can perform outside of their labs and controlled conditions, within a limited time, and in some cases adapt to changing conditions within a task. For this year’s event in Yokohama, RGMC organized five tracks grouped into two categories: the manufacturing track and four essential skill tracks. The essential skill tracks include cloth manipulation, in-hand manipulation, picking in clutter, and human-to-robot handovers. All tasks in the tracks are performed autonomously by the robot, without any human intervention. The tasks in all the tracks were co-developed based on the benchmarks published in the literature by the organizers and the benchmarks’ authors. This year’s RGMC attracted the interest of over 70 teams worldwide across the various competition tracks. Among the registered teams and through a reviewing process, 31 teams from 10 different countries were selected to participate in Yokohama. Only a fraction of these qualified teams could commit and complete the tracks on site, with 11 teams competing for the cloth manipulation track. Only 20 teams could commit and complete the tracks on-site, with three teams participating in three tracks and two teams participating each in two tracks.

The contactless estimation of the weight of a container and the amount of its content manipulated by a person are key pre-requisites for safe human-to-robot handovers. However, opaqueness and transparencies of the container and the content, and variability of materials, shapes, and sizes, make this problem challenging. In this paper, we present a range of methods and an open framework to benchmark acoustic and visual perception for the estimation of the capacity of a container, and the type, mass, and amount of its content. The framework includes a dataset, specific tasks and performance measures. We conduct a fair and in-depth comparative analysis of methods that used this framework and audio-only or vision-only baselines designed from related works. Based on this analysis, we can conclude that audio-only and audio-visual classifiers are suitable for the estimation of the type and amount of the content using different types of convolutional neural networks, combined with either recurrent neural networks or a majority voting strategy, whereas computer vision methods are suitable to determine the capacity of the container using regression and geometric approaches. Classifying the content type and level using only audio achieves a weighted average F1-score up to 81% and 97%, respectively. Estimating the container capacity with vision-only approaches and filling mass with audio-visual approaches, multi-stage algorithms reaches up to 65% weighted average capacity and mass scores. These results show that there is still room of improvement for the design of future methods that will be ranked and compared on the individual leaderboards provided by our open framework.

Binary descriptors are widely used for multi-view matching and robotic navigation. However, their matching performance decreases considerably under severe scale and viewpoint changes in non-planar scenes. To overcome this problem, we propose to encode the varying appearance of selected 3D scene points tracked by a moving camera with compact spatio-temporal descriptors. To this end, we first track interest points and capture their temporal variations at multiple scales. Then, we validate feature tracks through 3D reconstruction and compress the temporal sequence of descriptors by encoding the most frequent and stable binary values. Finally, we determine multi-scale correspondences across views with a matching strategy that handles severe scale differences. The proposed spatio-temporal multi-scale approach is generic and can be used with a variety of binary descriptors. We show the effectiveness of the joint multi-scale extraction and temporal reduction through comparisons of different temporal reduction strategies and the application to several binary descriptors.

The real-time estimation through vision of the physical properties of objects manipulated by humans is important to inform the control of robots for performing accurate and safe grasps of objects handed over by humans. However, estimating the 3D pose and dimensions of previously unseen objects using only RGB cameras is challenging due to illumination variations, reflective surfaces, transparencies, and occlusions caused both by the human and the robot. In this paper we present a benchmark for dynamic human-to-robot handovers that do not rely on a motion capture system, markers, or prior knowledge of specific objects. To facilitate comparisons, the benchmark focuses on cups with different levels of transparencies and with an unknown amount of an unknown filling. The performance scores assess the overall system as well as its components in order to help isolate modules of the pipeline that need improvements. In addition to the task description and the performance scores, we also present and distribute as open source a baseline implementation for the overall pipeline to enable comparisons and facilitate progress.

Object tracking is an ubiquitous problem that appears in many applications such as remote sensing, audio processing, computer vision, human-machine interfaces, human-robot interaction, etc. Although thoroughly investigated in computer vision, tracking a time-varying number of persons remains a challenging open problem. In this paper, we propose an on-line variational Bayesian model for multi-person tracking from cluttered visual observations provided by person detectors. The paper has the following contributions. We propose a variational Bayesian framework for tracking an unknown and varying number of persons. Our model results in a variational expectation-maximization (VEM) algorithm with closed-form expressions both for the posterior distributions of the latent variables and for the estimation of the model parameters. The proposed model exploits observations from multiple detectors, and it is therefore multimodal by nature. Finally, we propose to embed both object-birth and object-visibility processes in an effort to robustly handle temporal appearances and disappearances. Evaluated on classical multiple person tracking datasets, our method shows competitive results with respect to state-of-the-art multiple-object tracking algorithms, such as the probability hypothesis density (PHD) filter, among others.

Visual affordance segmentation identifies image regions of an object an agent can interact with. Existing methods re-use and adapt learning-based architectures for semantic segmentation to the affordance segmentation task and evaluate on small-size datasets. However, experimental setups are often not reproducible, thus leading to unfair and inconsistent comparisons. In this work, we benchmark these methods under a reproducible setup on two single objects scenarios, tabletop without occlusions and hand-held containers, to facilitate future comparisons. We include a version of a recent architecture, Mask2Former, re-trained for affordance segmentation and show that this model is the best-performing on most testing sets of both scenarios. Our analysis show that models are not robust to scale variations when object resolutions differ from those in the training set.

Accurately predicting whether an image is private before sharing it online is difficult due to the vast variety of content and the subjective nature of privacy itself. In this paper, we evaluate privacy models that use objects extracted from an image to determine why the image is predicted as private. To explain the decision of these models, we use feature-attribution to identify and quantify which objects (and which of their features) are more relevant to privacy classification with respect to a reference input (i.e., no objects localised in an image) predicted as public. We show that the presence of the person category and its cardinality is the main factor for the privacy decision. Therefore, these models mostly fail to identify private images depicting documents with sensitive data, vehicle ownership, and internet activity, or public images with people (e.g., an outdoor concert or people walking in a public space next to a famous landmark). As baselines for future benchmarks, we also devise two strategies that are based on the person presence and cardinality and achieve comparable classification performance of the privacy models.

Visual affordance segmentation identifies the surfaces of an object an agent can interact with. Common challenges for the identification of affordances are the variety of the geometry and physical properties of these surfaces as well as occlusions. In this paper, we focus on occlusions of an object that is hand-held by a person manipulating it. To address this challenge, we propose an affordance segmentation model that uses auxiliary branches to process the object and hand regions separately. The proposed model learns affordance features under hand-occlusion by weighting the feature map through hand and object segmentation. To train the model, we annotated the visual affordances of an existing dataset with mixed-reality images of hand-held containers in third-person (exocentric) images. Experiments on both real and mixed-reality images show that our model achieves better affordance segmentation and generalisation than existing models.

We propose a decentralised view-overlap recognition framework that operates across freely moving cameras without the need of a reference 3D map. Each camera independently extracts, aggregates into a hierarchical structure, and shares feature-point descriptors over time. A view overlap is recognised by view-matching and geometric validation to discard wrongly matched views. The proposed framework is generic and can be used with different descriptors. We conduct the experiments on publicly available sequences as well as new sequences we collected with hand-held cameras. We show that Oriented FAST and Rotated BRIEF (ORB) features with Bags of Binary Words within the proposed framework lead to higher precision and a higher or similar accuracy compared to NetVLAD, RootSIFT, and SuperGlue.

Jointly estimating hand and object shape ensures the success of the robot grasp in human-to-robot handovers. However, relying on hand-crafted prior knowledge about the geometric structure of the object fails when generalising to unseen objects, and depth sensors fail to detect transparent objects such as drinking glasses. In this work, we propose a stereo-based method for hand-object reconstruction that combines single-view reconstructions probabilistically to form a coherent stereo reconstruction. We learn 3D shape priors from a large synthetic hand-object dataset to ensure that our method is generalisable, and use RGB inputs instead of depth as RGB can better capture transparent objects. We show that our method achieves a lower object Chamfer distance compared to existing RGB based hand-object reconstruction methods on single view and stereo settings. We process the reconstructed hand-object shape with a projection-based outlier removal step and use the output to guide a human-to-robot handover pipeline with wide-baseline stereo RGB cameras. Our hand-object reconstruction enables a robot to successfully receive a diverse range of household objects from the human.

Estimating the 6D pose and size of household containers is challenging due to large intra-class variations in the object properties, such as shape, size, appearance, and transparency. The task is made more difficult when these objects are held and manipulated by a person due to varying degrees of hand occlusions caused by the type of grasps and by the viewpoint of the camera observing the person holding the object. In this paper, we present a mixed-reality dataset of hand-occluded containers for category-level 6D object pose and size estimation. The dataset consists of 138,240 images of rendered hands and forearms holding 48 synthetic objects, split into 3 grasp categories over 30 real backgrounds. We re-train and test an existing model for 6D object pose estimation on our mixed-reality dataset. We discuss the impact of the use of this dataset in improving the task of 6D pose and size estimation.

Matching views across independently moving cameras is important for 3D reconstruction and navigation. However, severe geometric and photometric differences, such as viewpoint, scale, and illumination changes, can considerably decrease the matching performance. This thesis proposes novel, compact, local features that can cope with scale and viewpoint variations. We extract and describe an image patch at different scales of an image pyramid by comparing intensity values between learned pixel-pair locations, and employ a cross-scale distance when matching these features. We capture, at multiple scales, the temporal changes of a 3D point, as observed in the visual stream of a camera, by tracking local image binary descriptors. After validating the feature-point trajectories through 3D reconstruction, we reduce, for each scale, the sequence of binary descriptors to a compact, fixed-length descriptor that identifies the most frequent and the most stable binary tests over time. We organise local spatio-temporal features extracted at a single scale in an incremental and adaptive ternary tree, which is stored locally within each camera. We then propose a novel decentralised approach that performs online cross-camera visual place recognition simultaneously to the formation of the tree, while the camera processes the video stream. Experiments show that the proposed approaches improve the accuracy when matching features across cameras, compared to existing approaches.