Empirically Compensated Setpoint Tracking for Spherical Robots With Pressurized Soft-Shells

Abstract

Replacing spherical robots' hard shells with soft, pressurized tires has the potential to improve their off-road practicality immensely. This change leverages spherical robots as a simple and rugged solution to problems currently addressed using wheeled or tracked vehicles. Though numerous prototypes have been launched over the last three decades, there has not been a spherical robot that poses a serious contender to tracked and wheeled systems. Most prototypes are built with a hard outer shell for ease of construction and control. Hard outer shells fail to absorb the impacts from uneven terrain. We addressed this issue by constructing a one-of-a-kind spherical robot with a durable pneumatic, soft outer shell. Although a soft-shell is more desirable for locomotion, it introduces complicated, nonlinear shell dynamics that cause a more challenging control problem. This article presents an empirical model of the steady-state torque induced by soft-shell dynamics, developed using system identification and a model based on tire dynamics. We show how this model, which fingerprints the robot's contact dynamics, is incorporated into RoboBall's steering control algorithm to compensate for soft-shell effects, enhancing setpoint tracking and improving control performance.