TI-P: Tactile-based Interactive Motion Planner in Unknow Cluttered Environemnts

Chengjin wang, zheng yan, Yanmin Zhou, Zhipeng Wang, Runjie Shen, Bin He

Keywords: interactive motion planning, tactile perception, unknown cluttered environments, perception-motion closed loop

TL;DR: This paper presents a tactile-based interactive planner. Robots can autonomously perceive operable obstacles and reconfigure spatial state to overcomes light suffer and unsolvable free-motion space challenges posed by unknown cluttered environments.

Abstract: Robotic motion in unknown cluttered environments often failures from catastrophic collisions and obstructions due to constrained free-motion space and light-suffer challenges. Multimodal tactile perception with force and proximity sensing offers inherent advantages in overcoming these limitations. This paper proposes a tactile-based interactive motion planning method (TI-P) using multimodal tactile sensing, which utilizes real-time tactile feedback to perceive the environment and infer the force-displacement characteristics of interacting objects. These interaction features are integrated into a sampling-based motion planner to predict the maximum connectivity probability of candidate trajectories. Subsequently, the planner interpolates sampled points and extrapolates the motion of objects along the trajectory to compute the optimal interaction forces for driving the robot. Simulation results demonstrate that the proposed planner effectively guides the robot to compliantly manipulate obstacles in its path, significantly improving motion adaptability in unknown cluttered environments.

Submission Number: 4