4 July 2018 in IEEE Robotics and Automation Letters
A deep learning architecture is proposed to predict graspable locations for robotic manipulation. It considers situations where no, one, or multiple object(s) are seen. By defining the learning problem to be classified with null hypothesis competition instead of regression, the deep neural network with red, green, blue and depth (RGB-D) image input predicts multiple grasp candidates for a single object or multiple objects, in a single shot. The method outperforms state-of-the-art approaches on the Cornell dataset with 96.0% and 96.1% accuracy on imagewise and object-wise splits, respectively. Evaluation on a multiobject dataset illustrates the generalization capability of the architecture. Grasping experiments achieve 96.0% grasp localization and 89.0% grasping success rates on a test set of household objects. The real-time process takes less than 0.25 s from image to plan.
26 February 2021 in IEEE Robotics and Automation Letters
This letter investigates the addition of keypoint detections to a deep network affordance segmentation pipeline. The intent is to better interpret the functionality of object parts from a manipulation perspective. While affordance segmentation does provide label information about the potential use of object parts, it lacks predictions on the physical geometry that would support such use. The keypoints remedy the situation by providing structured predictions regarding position, direction, and extent. To support joint training of affordances and keypoints, a new dataset is created based on the UMD dataset. Called the UMD+GT affordance dataset, it emphasizes household objects and affordances. The dataset has a uniform representation for five keypoints that encodes information about where and how to manipulate the associated affordance. Visual processing benchmarking shows that the trained network, called AffKp, achieves the state-of-the-art performance on affordance segmentation and satisfactory result on keypoint detection. Manipulation experiments show more stable detection of the operating position for AffKp versus segmentation-only methods and the ability to infer object part pose and operating direction for task execution.
28 February 2022 in The International Journal of Robotics Research
Contemporary grasp detection approaches employ deep learning to achieve robustness to sensor and object model uncertainty. The two dominant approaches design either grasp-quality scoring or anchor-based grasp recognition networks. This paper presents a different approach to grasp detection by treating it as keypoint detection in image-space. The deep network detects each grasp candidate as a pair of keypoints, convertible to the grasp representation g = { x, y, w, θ} T, rather than a triplet or quartet of corner points. Decreasing the detection difficulty by grouping keypoints into pairs boosts performance. To promote capturing dependencies between keypoints, a non-local module is incorporated into the network design. A final filtering strategy based on discrete and continuous orientation prediction removes false correspondences and further improves grasp detection performance. GKNet, the approach presented here, achieves a good balance between accuracy and speed on the Cornell and the abridged Jacquard datasets (96.9% and 98.39% at 41.67 and 23.26 fps). Follow-up experiments on a manipulator evaluate GKNet using four types of grasping experiments reflecting different nuisance sources: static grasping, dynamic grasping, grasping at varied camera angles, and bin picking. GKNet outperforms reference baselines in static and dynamic grasping experiments while showing robustness to varied camera viewpoints and moderate clutter. The results confirm the hypothesis that grasp keypoints are an effective output representation for deep grasp networks that provide robustness to expected nuisance factors.
17 February 2025 in IEEE Transactions on Automation Science and Engineering
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 at https://sites.google.com/view/foundationgrasp.
23 July 2019 in IEEE Robotics and Automation Letters
This paper presents a framework to detect and rank affordances of novel objects to assist with robotic manipulation tasks. The framework segments the affordance map of unseen objects using region-based affordance segmentation. Detected affordances define an initial state from which to generate action primitives for manipulation via the Planning Domain Definition Language (PDDL). The proposed category-agnostic affordance segmentation approach generalizes learned affordances to unseen objects by utilizing binary classification on proposed instance masks. The predicted pixel-wise level affordances are ranked by KL-divergence, augmenting the available affordance choices for manipulation tasks with non-primary affordances of an object. Experimental results show that the proposed method achieves state-of-the-art performance on affordance segmentation of novel objects, and outperforms baselines on affordance ranking. Real-world robotic manipulation demonstrates the use of affordance detection with PDDL-generated action primitives for task execution. Prediction of ranked affordances on unseen objects provides flexibility to accomplish goal-oriented tasks.
12 July 2022 in IEEE Robotics and Automation Letters
This paper investigates human instruction following for robotic manipulation via a hybrid, modular system with symbolic and connectionist elements. Symbolic methods build modular systems with semantic parsing and task planning modules for producing sequences of actions from natural language requests. Modern connectionist methods employ deep neural networks that learn visual and linguistic features for mapping inputs to a sequence of low-level actions, in an end-to-end fashion. The hybrid, modular system blends these two approaches to create a modular framework: it formulates instruction following as symbolic goal learning via deep neural networks followed by task planning via symbolic planners. Connectionist and symbolic modules are bridged with Planning Domain Definition Language. The vision-and-language learning network predicts its goal representation, which is sent to a planner for producing a task-completing action sequence. For improving the flexibility of natural language, we further incorporate implicit human intents with explicit human instructions. To learn generic features for vision and language, we propose to separately pretrain vision and language encoders on scene graph parsing and semantic textual similarity tasks. Benchmarking evaluates the impacts of different components of, or options for, the vision-and-language learning model and shows the effectiveness of pretraining strategies. Manipulation experiments conducted in the simulator AI2THOR show the robustness of the framework to novel scenarios.