Trajectory tracking of unmanned tracked vehicle based on model-free algorithm for off-road driving conditions

For unmanned tracked vehicles (UTVs) driving under off-road conditions, establishing an accurate model for trajectory tracking can be difficult mainly owing to the complex terrain-track interactions. Moreover, higher accuracy increases the computational complexity and convergence time. To solve this conundrum, this paper proposes a novel tracked vehicle tracking control method called the model-free tracking algorithm (MFTA), which combines model-free adaptive control theory with the traditional trajectory tracking control system of an UTV. Compared with the existing model-based trajectory tracking methods, the proposed MFTA does not rely on the vehicle model but uses the end-to-end data to complete trajectory tracking of the UTV. It can improve the generalization performance of the algorithm and solve the problem of vehicle parameter difference. Both simulations and real vehicle tests were carried out. The results show that the new MFTA can effectively complete trajectory tracking tasks while greatly reducing computational cost, which is an important indicator of improvement for trajectory tracking algorithms.