Trajectory optimization for asteroid landing with two-phase free final time

This work develops an autonomous trajectory planning algorithm for 6-DOF asteroid landing. The trajectory planning problem is formulated as a nonconvex time-optimal optimization problem with two-phase free final time, while the cost is regularized by augmenting a fuel consumption penalty. The nonconvex optimization problem is solved in successive solution method, and successive convexification is used to convert the original nonconvex problem into a sequence of convex subproblems, where each subproblem is obtained by linearizing the nonconvex dynamics and state constraints and using the velocity increment to give a convex expression of the fuel consumption penalty in cost function. Specifically, in the linearization, we divide the flight time interval into two parts and normalize each part using a time dilation coefficient to solve the problem that both the final times for the two flight phases are unknown, so that the original free final time problem turns to a fixed-time problem by minimizing the sum of the two time dilation coefficients and fuel consumption penalty. Besides, trust regions and virtual control are used to increase robustness of the algorithm. A convergence analysis is presented which indicates the successive solution will recover the local optimality of the original problem. Then the validity of the proposed algorithm and effects of different factors on flight time and fuel consumption are examined by simulations of landing on an irregular asteroid.