Mars entry guidance law design with neural network based HJB approach

In this paper, we address the Mars entry guidance problem by proposing a Hamilton-Jacobi-Bellman (HJB) approach based on neural network (NN). Initially, by considering initial, process and final constraints, a nominal feasible trajectory is computed using sequential quadratic programming. Subsequently, based on the nominal trajectory, the Mars entry guidance problem is transformed into a fixed-time optimal tracking control problem, which is equivalent to solving a HJB equation. However, the HJB equation is a nonlinear partial differential equation that has proven to be impossible to solve analytically. Thus, a NN based method is proposed to solve the HJB equation approximately. Finally, the developed method is used for the Mars entry guidance law design, and the simulation results demonstrate its effectiveness.