ANTAGONISTIC PUMP WITH MULTIPLE PUMPING MODES FOR ON-DEMAND SOFT ROBOT ACTUATION AND CONTROL

Abstract

Pneumatic soft robots are prized for their flexibility in achieving adaptable deformations and compliance adjustments. However, the conventional pumps used in these systems often rely on simplistic, sensorless, and ON–OFF control mechanisms, which limit the potential of these robots. Drawing inspiration from the intricate functionality of the natural heart-pumping mechanism, we present an innovative and versatile pump that integrates an antagonistic pumping mechanism with a reinforcement learning-powered control strategy. The antagonistic pump features dual chambers for expansion and deflation, valve-controlled interconnections, and distributed pressure sensors. This dedicated architecture enables a nuanced air exchange logic and pumping sequence, thereby facilitating a wide range of pneumatic actuation possibilities for pneumatic soft robots. Our pressure control method leverages the structural capabilities and degrees of freedom inherent in the pump, enhancing pumping efficiency and precision. Diverging from the traditional controllers, it autonomously evaluates the properties of unknown connected loads and dynamically adjusts pumping actions accordingly. Consequently, the pump provides multiple pumping modes, including load perception, rapid inflation, dual-load control, adaptive pumping across positive and negative pressure ranges, fine-tuning, and instantaneous pressure switching. Moreover, by iteratively executing the pumping cycle, the pump can extend its output pressure limits. We have successfully built and tested a prototype pump, validating its ability to achieve a broad range of pressures with precise and robust control. These results underscore the pump’s potential to actuate diverse soft robots through multimode pumping, offering a pioneering solution for universal soft robot actuation and control.