MPC-Based Position Control for Multi-Segment Pneumatic Soft Manipulator Considering Obstacle Avoidance

Multi-segment pneumatic soft manipulators are highly valued for their predominated characteristics in safety and dexterity. However, achieving precise position control in the presence of obstacles remains challenging, particularly when targets are close to the obstacle. Model predictive control (MPC) offers a promising solution by modeling obstacle avoidance as system state constraints that can be efficiently addressed by the controller. This letter presents a double closed-loop control framework based on MPC, designed to enable precise position control while avoiding the obstacle for a multi-segment pneumatic soft manipulator. We develop and compare three distinct obstacle avoidance functions (OAFs) to determine the most effective one, which is integrated into the MPC scheme. After obtaining muscle lengths from the MPC, a feedback control adjusts the mapping between muscle length and pressure in real-time. The feasibility of the proposed MPC-based controller is validated through simulations using piecewise constant curvature (PCC) kinematics. Then, physical experiments verify the effectiveness and robustness of each OAF. The collision numbers calculated from the repetitive experiments indicate that the MPC-control barrier function (CBF) outperforms others in both position control with obstacle presence.