Sample-Based Passively Compliant Manipulation for Senserless Contact-Rich Scenarios with Uncertainties

摘要

In this paper, we present a novel incremental approach to addressing uncertainty in robotic manipulation tasks for contact-rich environments. Traditional methods in contact-rich environments often rely on continuous perception, force/torque sensors and high-frequency controllers. The key idea of our approach is leveraging the passive deformation of flexible components as a feedback mechanism to generate passively adaptable trajectories that can cope with uncertainty. Our algorithm focused on planning the high-level trajectory and introduced the path planning problem in the deviation-fusion process. Our approach doesn’t require high-frequency controllers, force/torque sensors, and high-precision simulation. Instead, we incorporate errors into uncertainties to ensure the validity of our common planned trajectories even if the environment’s state is not perfectly obtained.