Trajectory Tracking of Mobile Robots with Slip and Skid Compensation
Autonomous wheeled mobile robots are capable of working in outdoor environments to reduce the risk for human operators or increase productivity. Accounting for wheel-terrain interaction is crucial for the navigation and traction control of mobile robots in outdoor environments and rough terrains. Wheel-terrain interaction could affect the dynamics and controllability of the robot, which could cause slipping, skidding or loss of mechanical stability. In order to reduce risks related to wheel-terrain interaction, the effects of this interaction should be considered when designing the navigation system. However, the robot could experience different unanticipated surface types and terrain conditions while traversing in outdoor environments, which makes it hard or almost impossible to pre-model every significant detail. Hence, an online adaptive method is needed to capture wheel-terrain interaction effects on the robot’s behaviour in outdoor environments. The capacity of solving highly nonlinear problems has made artificial intelligence a common approach to be able to utilize and adapt robots to different situations in outdoor environments. Wheel slip is one of the surface hazards that need to be detected to mitigate the risk of losing the robot’s controllability or mission failure occurring. Despite prior research in this area, the open problems are (1) the need for in situ slip estimation in harsh environments using low-cost/power and easy to integrate sensors, and (2) removing the need for prior information about the soil, as this is not always available. This thesis addresses these open problems and presents a novel slip estimation method that utilizes only two proprioceptive sensors (IMU and wheel encoder) to estimate the slippage at the vehicle-level using deep learning methods. It is experimentally shown to be real-world feasible in outdoor, uneven terrains without prior soil information assumptions. Comparison with previously used machine learning algorithms for continuous and discrete slip estimation problems show more than 9% and 14% improvement in estimation performance respectively.
Skidding is another surface hazard for mobile robots’ navigation and traction control systems operating in outdoor environments and rough terrains due to the wheel-terrain interaction. It could lead to high trajectory tracking errors, losing the robot’s controllability, and mission failure occurring. As was the case for the slip estimation problem, despite research in this field, the development of a real-world feasible in situ skid estimation system with the capability of operating in harsh and unforeseen environments using low-cost/power and easy to integrate sensors is still an open problem in terramechanics. This thesis presents a novel velocity-based definition and a real-world feasible in situ estimation for the mobile robot’s undesired skidding at the vehicle-level in outdoor environments. The proposed technique estimates the undesired skidding using a combination of two proprioceptive sensors (e.g., IMU and wheel encoder) and deep learning to address the open problem in this domain. The practicality of a velocity-based definition and the performance of the proposed undesired skid estimation technique are evaluated experimentally. The results show that the proposed technique performs with less than 11.79 mm/sec mean absolute error and estimates the direction of undesired skidding with approximately 98% accuracy.
The final contribution of this work is to design a robust trajectory tracking system with both slip and skid compensation at the vehicle-level for outdoor environments. In addition to the general slipping and skidding hazards for mobile robots in outdoor environments, slip and skid cause uncertainty for the trajectory tracking system and put the validity of stability analysis at risk. Despite research in this field, having a real-world feasible online slip and skid compensator for outdoor and harsh environments is still an open problem. This thesis presents a novel trajectory tracking technique with real-world feasible online slip and skid compensation at the vehicle-level for skid-steering mobile robots in outdoor environments. Dealing with slip and skid at the vehicle-level has the advantage of only requiring two parameters rather than the conventional three slip and skid parameters at the wheel-level. The sliding-mode control is utilized as the trajectory tracking technique to steer the robot based on the robot’s kinematics and dynamics model. The developed deep learning models are integrated into the control-feedback loop to estimate the robot’s slipping and skidding and feed the compensator in a real-time manner. The experimental results show that the proposed controller with the slip and skid compensator improves the performance of the trajectory tracking system by more than 27%.