Task-oriented grasping (TOG), which refers to synthesizing grasps on an object that are configurationally compatible with the downstream manipulation task, is the first milestone towards tool manipulation. Analogous to the activation of two brain regions responsible for semantic and geometric reasoning during cognitive processes, modeling the intricate relationship between objects, tasks, and grasps necessitates rich semantic and geometric prior knowledge about these elements. Existing methods typically restrict the prior knowledge to a closed-set scope, limiting their generalization to novel objects and tasks out of the training set. To address such a limitation, we propose FoundationGrasp, a foundation model-based TOG framework that leverages the open-ended knowledge from foundation models to learn generalizable TOG skills. Extensive experiments are conducted on the contributed Language and Vision Augmented TaskGrasp (LaViA-TaskGrasp) dataset, demonstrating the superiority of FoundationGrasp over existing methods when generalizing to novel object instances, object classes, and tasks out of the training set. Furthermore, the effectiveness of FoundationGrasp is validated in real-robot grasping and manipulation experiments on a 7-DoF robotic arm. Our code, data, appendix, and video are publicly available athttps://sites.google.com/view/foundationgrasp. Note to Practitioners—This research is motivated by the challenge of generalizable task-oriented grasping skill learning. Solving such a challenge could significantly improve the robot’s level of automation and intelligence in tool manipulation for household and industrial tasks. Existing methods struggle with handling unseen objects and tasks in dynamic, open-world environments. To overcome this limitation, we propose to leverage the open-ended knowledge from foundation models to improve the generalization capabilities of existing TOG methods. This way, the robot can perform TOG w.r.t. unseen objects and tasks, facilitating downstream tool manipulation. Overall, this research has broad applicability to various scenarios involving tool manipulation, such as cleaning kitchenware and assembling parts in industrial contexts.

Deep neural networks are susceptible to generating overconfident yet erroneous predictions when presented with data beyond known concepts. This challenge underscores the importance of detecting out-of-distribution (OOD) samples in the open world. In this work, we propose a novel feature-space OOD detection score based on class-specific and class-agnostic information. Specifically, the approach utilizes Whitened Linear Discriminant Analysis to project features into two subspaces the discriminative and residual subspaces - for which the in-distribution (ID) classes are maximally separated and closely clustered, respectively. The OOD score is then determined by combining the deviation from the input data to the ID pattern in both subspaces. The efficacy of our method, named WDiscOOD, is verified on the large-scale ImageNet-1k benchmark, with six OOD datasets that cover a variety of distribution shifts. WDiscOOD demonstrates superior performance on deep classifiers with diverse backbone architectures, including CNN and vision transformer. Furthermore, we also show that WDiscOOD more effectively detects novel concepts in representation spaces trained with contrastive objectives, including supervised contrastive loss and multi-modality contrastive loss.

The success of 6-DoF grasp learning with point cloud input is tempered by the computational costs resulting from their unordered nature and pre-processing needs for reducing the point cloud to a manageable size. These properties lead to failure on small objects with low point cloud cardinality. Instead of point clouds, this manuscript explores grasp generation directly from the RGB-D image input. The approach, called Keypoint-GraspNet (KGN), operates in perception space by detecting projected gripper keypoints in the image, then recovering their SE(3) poses with a$\mathrm{P}n\mathrm{P}$algorithm. Training of the network involves a synthetic dataset derived from primitive shape objects with known continuous grasp families. Trained with only single-object synthetic data, Keypoint-GraspNet achieves superior result on our single-object dataset, comparable performance with state-of-art baselines on a multi-object test set, and outperforms the most competitive baseline on small objects. Keypoint-GraspNet is more than 3x faster than tested point cloud methods. Robot experiments show high success rate, demonstrating KGN's practical potential.

We propose an improved keypoint approach for 6-DoF grasp pose synthesis from RGB-D input. Keypoint-based grasp detection from image input demonstrated promising results in a previous study, where the visual information provided by color imagery compensates for noisy or imprecise depth measurements. However, it relies heavily on accurate keypoint prediction in image space. We devise a new grasp generation network that reduces the dependency on precise keypoint estimation. Given an RGB-D input, the network estimates both the grasp pose and the camera-grasp length scale. Re-design of the keypoint output space mitigates the impact of keypoint prediction noise on Perspective-n-Point (PnP) algorithm solutions. Experiments show that the proposed method outperforms the baseline by a large margin, validating its design. Though trained only on simple synthetic objects, our method demonstrates sim-to-real capacity through competitive results in real-world robot experiments.

We consider the task of grasping a target object based on a natural language command query. Previous work primarily focused on localizing the object given the query, which requires a separate grasp detection module to grasp it. The cascaded application of two pipelines incurs errors in overlapping multi-object cases due to ambiguity in the individal outputs. This work proposes a model named Command Grasping Network (CGNet) to directly output command satisficing grasps from RGB image and textual command inputs. A dataset with ground truth (image, command, grasps) tuple is generated based on the VMRD dataset to train the proposed network. Experimental results on the generated test set show that CGNet outperforms a cascaded object-retrieval and grasp detection baseline by a large margin. Three physical experiments demonstrate the functionality and performance of CGNet.

A human-in-the-loop system is proposed to enable hands-free collaborative manipulation for people with physical disabilities. Studies show that the cognitive burden of interfacing with a robotic assistant decreases with increased robot autonomy. Incorporating modern advances in perception with augmented reality, this paper describes a framework for obtaining high-level intents from the user to specify manipulation tasks for execution. Augmented reality glasses provide an egocentric perspective to the robot. The glasses also provide visual feedback to users on a virtual menu showing a summary of robot affordances. The system processes the vision input to interpret the users environment. A Tongue Drive System serves as the input modality for triggering task execution by the robotic arm. Several manipulation experiments are performed with comparison to Cartesian control. The outcomes are also compared to reported state-of-the-art approaches. The results demonstrate competitive performance with minimal user input requirements.