Abstract:There has been significant recent work on data-driven algorithms for learning general-purpose grasping policies. However, these policies can consistently fail to grasp challenging objects which are significantly out of the distribution of objects in the training data or which have very few high quality grasps. Motivated by such objects, we propose a novel problem setting, Exploratory Grasping, for efficiently discovering reliable grasps on an unknown polyhedral object via sequential grasping, releasing, and toppling. We formalize Exploratory Grasping as a Markov Decision Process, study the theoretical complexity of Exploratory Grasping in the context of reinforcement learning and present an efficient bandit-style algorithm, Bandits for Online Rapid Grasp Exploration Strategy (BORGES), which leverages the structure of the problem to efficiently discover high performing grasps for each object stable pose. BORGES can be used to complement any general-purpose grasping algorithm with any grasp modality (parallel-jaw, suction, multi-fingered, etc) to learn policies for objects in which they exhibit persistent failures. Simulation experiments suggest that BORGES can significantly outperform both general-purpose grasping pipelines and two other online learning algorithms and achieves performance within 5% of the optimal policy within 1000 and 8000 timesteps on average across 46 challenging objects from the Dex-Net adversarial and EGAD! object datasets, respectively. Initial physical experiments suggest that BORGES can improve grasp success rate by 45% over a Dex-Net baseline with just 200 grasp attempts in the real world. See <a class="link-external link-https" href="https://tinyurl.com/exp-grasping" rel="external noopener nofollow">this https URL</a> for supplementary material and videos.

Selection of Secure Gravity-Based Caging Grasps of Planar Objects: Robustness and Experimental Validation

Computation for Maximum Stable Grasping in Dynamic Force Distribution

Robust Adaptive Safe Robotic Grasping with Tactile Sensing

EfficientGrasp: A Unified Data-Efficient Learning to Grasp Method for Multi-Fingered Robot Hands

Safe Robotic Grasping: Minimum Impact-Force Grasp Selection

A Convex-Combinatorial Model for Planar Caging

A Novel Geometry-based Algorithm for Robust Grasping in Extreme Clutter Environment

GRaCE: Balancing Multiple Criteria to Achieve Stable, Collision-Free, and Functional Grasps

Grasp Detection for Assistive Robotic Manipulation

Probabilistic Closed-Loop Active Grasping

Graspness Discovery in Clutters for Fast and Accurate Grasp Detection

Grasp Failure Constraints for Fast and Reliable Pick-and-Place Using Multi-Suction-Cup Grippers

Efficient End-to-End Detection of 6-DoF Grasps for Robotic Bin Picking

AnyGrasp: Robust and Efficient Grasp Perception in Spatial and Temporal Domains

Gravity-aware Grasp Generation with Implicit Grasp Mode Selection for Underactuated Hands

GE-Grasp: Efficient Target-Oriented Grasping in Dense Clutter

An Integrated Multi-Sensor Network for Adaptive Grasping of Fragile Fruits: Design and Feasibility Tests

Toward An Analytic Theory of Intrinsic Robustness for Dexterous Grasping

Caging in Time: A Framework for Robust Object Manipulation under Uncertainties and Limited Robot Perception

Efficient and Robust Robotic Grasping in Cluttered Scenes: A Point Cloud-Based Approach with Heuristic Evaluation.

Exploratory Grasping: Asymptotically Optimal Algorithms for Grasping Challenging Polyhedral Objects