Path-Planning and Control of Micro-Rover Swarms For Planetary Exploration Missions.

Future surface exploration missions to the Moon and Mars are likely to include wheeled microrovers. The deployment of such multiple rovers will enable exploration of wider regions for more science return. However, co-ordinated operation of micro-rovers on unknown, sandy and rugged terrains is extremely challenging compared to operation of single, large wheeled rovers. This motivated research on micro-rover swarms for planetary exploration and it aims at advancing the state-of-the-art of wheeled micro-rover swarm missions through developing an innovative control and path planning algorithm for safe locomotion and navigation. Based on the literature review, backed by preliminary analysis, it is proposed to use Artificial Potential Field (APF) method for path planning and Sliding Mode Control (SMC) for traction control of the required multi-agent configuration. Initially the swarming behaviour of point mass systems is studied using the second-order nonlinear model. Aggregation towards goal point and formation to desired geometry is achieved using APF and SMC. In order to overcome the limitations of point mass model, a novel algorithm is developed based on the dynamic model for a four wheeled rover, incorporating the strengths of APF and SMC in the path planning and traction control respectively. Further, this algorithm is modified to incorporate a navigation function for collision avoidance with predefined obstacles. Simulations were executed on uncertain terrain using the characteristics of an in-house soil stimulant. A potential function with minimum number of local minima is chosen for implementation. Although certain artificial potential functions suffer from the local minimum problem, APF is still an efficient method for collision avoidance due to its computational efficiency. Simulation results prove that the proposed robust controller effectively maintains the slip rate of the wheel and avoids the excessive spin and sinkage of the driving wheel under challenging operating conditions. The results have shown that the novel algorithm for wheeled robot control and path-planning have been able to avoid pre-defined obstacles and collisions in an unstructured environment like the Martian surface. The obstacles are assumed to be captured by LIDAR sensors for purpose of simulation. For planetary surface exploration missions, this will allow for faster sample acquisition and return for ground experimentation and research, achievable at a lower cost as compared to previous missions