Comparison of Semi-autonomous Mobile Robot Control Strategies in Presence of Large Delay Fluctuation

Abstract

We propose semi-autonomous control strategies to assist in the teleoperation of mobile robots under unstable communication conditions. A short-term autonomous control system is the assistance in the semi-autonomous control strategies, when the teleoperation is compromised. The short-term autonomous control comprises of lateral and longitudinal functions. The lateral control is based on an artificial potential field method where obstacles are repulsive, and a route is attractive. LiDAR-based artificial potential field methods are well studied. We present a novel artificial potential field method based on color and depth images. Benefit of a camera system compared to a LiDAR is that a camera detects color, is cheaper, and does not have moving parts. Moreover, utilization of active sensors is not desired in the particle accelerator environment. A set of experiments with a robot prototype are carried out to validate this system. The experiments are carried out in an environment which mimics the accelerator tunnel environment. The difficulty of the teleoperation is altered with obstacles. Fully manual and autonomous control are compared with the proposed semi-autonomous control strategies. The results show that the teleoperation is improved with autonomous, delay-dependent, and control-dependent assist compared to the fully manual control. Based on the operation time, control-dependent assist performed the best, reducing the time by 12% on the tunnel section with most obstacles. The presented system can be easily applied to common industrial robots operating e.g. in warehouses or factories due to hardware simplicity and light computational demand.