A collaborative path planning approach for multiple robots persistently building a lunar base

Abstract

The construction of lunar bases is a critical part of the in-depth implementation of lunar exploration missions, which poses great challenges due to the complex environment of the lunar surface and the high consumption of human resources. It is highly promising to employ multiple autonomous robots as the main body to execute construction-related complex tasks, such as inspection, transportation, building and so on, which requires the collaborative path planning for the robot team to meet the inherent temporal requirements of those tasks. This paper develops a controller synthesis method to plan obstacle-avoidance paths that not only satisfies temporal constraints of construction tasks, but also ensures the long-term autonomy of each robot in the team from the perspective of energy consumption. Firstly, linear temporal logic and model checking tools are used to generate reachability sequences that satisfy the specification regarding to the robot team’s global task. Secondly, this reachability sequence is formulated as a set of quadratic programming problems. By encoding the safety and reachability constraints into the controller through the control barrier function, trajectories that are both safe and satisfying temporal constraints are planned for multiple robots. In addition, our controller synthesis approach can also successfully solve the path planning problem of multiple robots subject to survival constraints. By the dedicated design of the energy-constrained barrier functions, we obtain a control strategy that guarantees long-term autonomy. Finally, in our simulation four robots are employed to accomplish the lunar base construction tasks described through LTL in two different obstacle environments, and the team performs the task for more than half an hour where every fully-charged robot can only work about forty-five seconds. The trajectories in the simulation results verify the feasibility and effectiveness of our proposed methods.