Dynamic Task Space Control of Redundant Pneumatically Actuated Soft Robot

Soft robotic manipulators promise manipulation dexterity and compliance, but these properties make them hard to control due to the uncertainties and nonlinearities in their dynamics. Towards solving these challenges, this paper proposes a real-time dynamic control approach for a pneumatically actuated redundant soft robotic arm. We first present a computationally-efficient dynamic model designed for real-time control. Next, we introduce a robust adaptive passivity-based control approach that complements the dynamic model to enhance the system's tracking performance under uncertainty. System uncertainties are categorized based on the bounds and behavior of dynamic terms in the model, with dominant stiffness and damping effects compensated by adaptive terms and unmodeled dynamics handled through the robust term. We first develop a control strategy in actuator space, then extend it to task space to encompass position control, orientation control, and simultaneous position and orientation control. To improve tracking performance in task space, we employ a redundancy resolution algorithm, addressing secondary tasks such as maintaining actuators within feasible limits and promoting homogeneous deformations in the robot. We comprehensively assess the performance of the proposed robust adaptive passivity control under various operation speeds, loading conditions, and trajectories. A comparison with recent baselines from the literature highlights how our method achieves both superior tracking performance and robot operation speed.