Evaluation of Two Path Following Controllers for an Ackermann Off-road Vehicle in Winter and Summer Conditions

Off-road driving can be a challenging task with rapid changes in the driving conditions, terrain and vehicle behavior. For off-road autonomous vehicles, it is important to be robust to these changes, and parts of this robustness comes from the path following controller of the vehicle. In this paper we compare two different path following controllers using a kinematic model of Ackermann vehicles, the Stanley controller and a controller originally made for unicycles, but adapted for Ackermann vehicles. The comparison is done using an Off-road Polaris vehicle modified for autonomous driving. Two experiments are conducted, where the first experiment is during wintertime, driving with belts in snowy conditions, and the second experiments is during summer time, driving with wheels in muddy conditions. Our research vehicle can be fitted with different payloads that affect the weight and center of gravity. Therefore our main motivation for using kinematic controllers is to have a simple and robust controller that handles all situations using the same control parameters. Both controllers are used on the same path, and the cross-track errors are compared. The main conclusion is that the Stanley controller showed the most robustness against variations in driving conditions and vehicle dynamics, compared to the adapted unicycle method.