Successive Convexification for Asteroid 6-DoF Powered Descent with PINN Gravity Model

In this paper, regarding the asteroid powered de-scent scenario, the asteroid gravity PINN model and the successive convexification method are adopted to solve the fuel-consumption minimal 6-DoF descent trajectory optimization problem, which can better ensure the feasibility of trajectory optimization problems. The dynamics and kinematics models are established for the probe's descent process. Due to the rotation of the asteroid and which cannot be ignored, both dynamics and kinematics models are relative motion models. A new PINN model has been established for asteroid gravity acceleration modeling, which is based on traditional DNN models and ac-tual physical laws are incorporated to improve the prediction generalization ability for unknown regions. For the trajectory optimization problem with minimum fuel consumption target, a possible thruster configuration is assumed and the successive convexification method is used to solve the problem with strong nonlinearity efficiently. Compared with the previous method, the method adopted in this paper can reduce asteroid gravity modeling errors by about 50%, and the performance of the optimized descent trajectory can also be improved by about 20%.