Heuristic Task Allocation Method for Heterogeneous Lunar Robots under Dynamic Resource Cost

China, in collaboration with multiple countries, is preparing to construct a long-term, independently operated international lunar research station on the surface of the moon in the coming years. The construction of an unmanned lunar research station requires the collaboration of diverse and heterogeneous lunar robots to successfully carry out missions. Task allocation for heterogeneous lunar robots is a crucial technical requirement, focusing on constructing evaluation matrix, designing optimization objective functions, and developing algorithms for optimization matching. The state changes during the execution of tasks by robots result in different resource costs for completing other task goals compared to the initial resource costs. Therefore, one of the major challenges in task allocation for multi-robot systems is effectively managing dynamic resource costs during the construction of evaluation matrix (and subsequent optimization based on the matrix). However, the current approach tends to calculate static cost without considering cost changes during task execution, which deviates from standard engineering practices. Instead, the primary contribution of this paper lies in designing a heuristic algorithm capable of optimizing task allocation outcomes while accounting for the dynamic changes in resource costs incurred during goal achievement. In this paper, we propose a heuristic task allocation method for heterogeneous lunar robots under dynamic resource cost. Given a multi-robot cooperative task, our method firstly constructs a task allocation model based on robotic capability and task resource consumption. Next, based on relaxation planning graph, the resource cost for each lunar rover to complete all task goals is calculated, and the cost of transitioning between different goal states is estimated (to efficiently update the resource cost of completing each goal as the lunar rover's state changes). Subsequently, the evaluation matrix is constructed. Then, a heuristic algorithm is developed based on greedy search to efficiently solve the problem of allocating tasks to multiple heterogeneous lunar robots. We perform extensive sets of experiments in relative domains and the results show the effectiveness of the proposed method.