Feedback Control of a Dielectric-Elastomer Soft Robot Using Modified Dubins Path Planning

Soft robots demonstrate enhanced safety and adaptability in complex and dynamic environments compared to rigid robots. Their unique actuation mechanisms, materials, structural designs, and environmental interactions necessitate distinct path planning methods. In this work, we develop a modified Dubins path planning method for a dielectric elastomer (DE) soft robot, integrating vision-based feedback control. Due to its structural design, the soft robot exhibits instantaneous forward, backward, and right-turn motions under different frequencies with a single voltage input. The modified Dubins path planning method accounts for these three distinct locomotion modes, providing optimal three- and two-segment paths. Image-based visual servoing enables real-time adjustments of the optimal path, enhancing adaptability under external disturbances. Experiments demonstrate that the soft robots can be controlled to reach targets from different initial orientations, navigate mazes, and achieve destinations despite wind disturbances.