Recently, there has been a surge of interest in achieving onboard real-time computation of optimal trajectories for endo-atmospheric launch vehicles. Despite their robustness, the efficiency of widespread techniques such as direct collocation methods strongly depends on computational capabilities, which often obliges to use them offline uniquely. Key candidates for achieving real-time operation with low computational load are shooting methods. Nevertheless, with the aim of ensuring a detailed and refined representation of the physics of the problem to prevent missions from failing, various state constraints and delays have to be considered, which further complexify the use of such procedures. In this talk, we show how to exploit geometric control to efficiently implement shooting methods for the optimal control of endo-atmospheric launch vehicles. As natural extension, we show how to adapt these techniques to design fast trajectory optimization algorithms for dynamical path-planning in cluttered environments.
Bienvenue à tous!