Effect of Single Versus Dual Wheel Drive on Ackermann Trajectory Deviation of a Decontamination Mobile Robot in Webots

Abstract

This paper evaluates the effect of rear-wheel drive configurations on the three-marker path deviation of an Ackermann-steered mobile robot. The study uses a Webots simulation and a physical prototype. The platform is developed as a service robot for decontamination tasks in nuclear hot-chamber and decommissioning environments, where precise path following is a prerequisite for safe remote operation. Two open-loop Webots scenarios are compared at a fixed rear steering angle of 45 degrees and a commanded outer-wheel speed of 2 rad/s: (i) an equal-speed dual rear-wheel drive (both rear wheels are at 2 rad/s) and (ii) a single outer-rear-wheel drive (wheel 4 is at 2 rad/s; the inner wheel is free-rolling). Ideal circular paths with a centre-of-gravity turning radius of 0.389 m are used as a reference. Point-wise marker Euclidean errors at three markers yield mean errors of 0.013 m and 0.016 m for the two equal-speed dual-drive turns and 0.006 m for both single outer-drive turns, i.e., an approximate 60% reduction in the mean marker error under the tested Webots condition (from approximately 0.015 m to 0.006 m). Complementary three-marker RMSE values confirm the same ranking (Scenario 1 is approximately 0.013–0.017 m; Scenario 2 is approximately 0.007 m) but are not an integrated tracking error over the full path. In the single outer-rear-wheel Webots runs, the free-rolling inner wheel (wheel 3) averaged |ωinner| approximately 1.27 rad/s; this speed was then commanded on the physical robot together with |ωouter| = 2 rad/s, i.e., as differentiated dual rear-wheel drive rather than as single-wheel drive. Physical-robot trajectories obtained by fusing drive-wheel encoders with the onboard IMU of an Intel RealSense D435i preserve the same qualitative ranking (a mean marker error of approximately 0.021 m for equal-speed dual drive versus approximately 0.005 m for differentiated dual drive). These physical paths are based on encoder + IMU odometry rather than independent ground truth. The results show that commanding identical speeds to the inner and outer rear wheels is inconsistent with Ackermann kinematics and increases the three-marker path deviation under the tested conditions, whereas ICR-consistent inner/outer speed pairs better approximate instantaneous centre of rotation (ICR) kinematics.