Control of Two-Wheeled Robots in Low-Traction Environments

Abstract

Much research on two-wheeled robots has been completed in the past decade, but robot stability in low-traction environments has not yet been considered. This work demonstrates, in both simulation and experiment, that robot balance is compromised on low-traction surfaces with conventional balance controllers. The two-wheeled robot system model was separated into the motor, sensor and structural dynamics. A novel structural dynamics model accounted for wheel slip effects by introducing friction modelling. The conventional wheel slip model was used to simulate friction behaviour for two low-traction surfaces: the first with characteristics similar to ice and the second frictionless; the final test case was the no-slip constraint. An inertia or reaction wheel was introduced as an auxiliary actuator to the basic two-wheeled robot configuration, to quantify the effects on controller performance. Two reference tracking controllers were developed for low-traction surfaces, and compared to a conventional balance controller. Controller performance was simulated on the three friction surfaces, and was assessed in terms of four control objectives: robot stability, speed regulation, robustness to disturbances and energy consumption. The latest controller is shown to be capable of transitions between frictionless and no-slip surfaces, maintaining balance and attempting speed regulation where possible. An experimental test rig for approximating frictionless surfaces used bearings attached to the wheel axles to lift the drive wheels from the ground. The latest controller was retuned in experiment and was able to achieve balance and speed regulation on both the no-slip and frictionless surfaces. The drive wheels were found to be ineffective for balancing on frictionless surfaces for extended periods. Conversely, the reaction wheel balanced the robot indefinitely, therefore it is a viable option for balancing in low-traction environments.