Rapid Trajectory Optimization of Multi-Agent Hybrid Systems

Abstract

Multi-agent systems in aerospace applications correspond to coordinated control of multiple vehicles such as those arising in multiple entry-vehicle systems, air-traffic control and formation flying to name a few examples. In certain applications, the vehicles may be docked or separated in mid-flight resulting in switches in the dimension of the continuous-time state and control variables. Dimensional switching distinguishes such problems in a somewhat unique way when compared to other multi-agent systems such as mobile robots where docking may not be allowed. A hybrid optimal control framework is presented that mathematically formalizes such mission planning problems towards the goal of rapid flight-plan-analyses via fast trajectory optimization. One motivation for this formalism is that many future systems are conceived as multi-agent systems and it has become increasingly apparent that a new framework is necessary to efficiently analyze various scenarios. The pseudospectral knotting method is proposed as a fast computational technique to solve such hybrid optimal control problems. The concepts are illustrated for a two-agent benchmark problem associated with a multi-agent launch system where the return vehicle is constrained by a return point requirement.