Motion planning of multiple unmanned vehicles using sequential convex programming-based distributed model predictive control

This article tackles the complex challenge of multi-vehicle motion planning by presenting the novel Sequential Convex Programming-based Distributed Model Predictive Control (SCP-DMPC) algorithm. Existing methods often struggle with the nonconvex nature of optimization in multi-vehicle motion planning, frequently compromising on computational efficiency. SCP-DMPC innovatively combines the predictive control of DMPC with the optimization prowess of SCP to address these issues efficiently, while adhering to both physical and operational constraints. To ensure precise attainment of target poses, a three-stage control strategy is introduced, with theoretical analysis on its recursive feasibility and asymptotic stability, enhancing the reliability of the algorithm. Moreover, a heuristic-based deadlock resolution scheme is devised to prevent vehicle stalling, a common issue in cooperative motion. Validated through simulations in challenging scenarios, including symmetric position swapping and obstacle-laden formation transition, SCP-DMPC demonstrates superior adaptability, precision, and efficiency. These results underscore its potential for robust, real-time unmanned vehicle applications, suggesting new directions for future research and development in the field.