Non Prehensile Manipulation

Non-prehensile manipulation (NPM) focuses on robotic manipulation techniques that do not involve grasping, instead utilizing pushing, sliding, or other contact-based interactions to move objects. Current research emphasizes robust control strategies in uncertain environments, often employing model predictive control (MPC), reinforcement learning (RL), and physics-based simulation to address challenges like object instability and contact dynamics. These advancements are crucial for improving robotic dexterity in cluttered spaces and enabling tasks such as object rearrangement, assembly at micro-scales, and manipulation of objects exceeding a robot's payload capacity, impacting fields ranging from manufacturing to minimally invasive surgery.