Non-Prehensile Manipulation
Overview
Non-prehensile manipulation studies how robots move objects through contact without enclosing them in a grasp, enabling tasks such as precise pushing, rearrangement, transport, and throwing. Approaches include model-based planning and model predictive control, reinforcement learning, learned dynamics, and geometric or graph-based planners that combine object-level goals with feasible robot motions. Work spans tabletop pushing and stacking as well as bimanual manipulation, mobile robots, and dynamic contact tasks, with simulation-to-real transfer and hardware demonstrations increasingly central. A key challenge is making plans reliable despite uncertain friction, object properties, and contact changes while keeping motions physically executable.
Latest papers
19 papers0 papers in the last four weeks, down 100% on the four weeks before. 0.0% of all new papers.