Predictor-searcher guidance for powered descent under multiple constraints

A novel guidance method called Predictor-Searcher Guidance (PSG) is proposed for the multi-constrained powered descent phase of a landing mission. The idea of the PSG method is to predict the state of the probe in a short prediction horizon and search the guidance database for the optimal control corresponding to the desired state. The dynamics of the probe is decoupled into the controllable dynamics and the uncontrollable dynamics, and the former is used to generate guidance database satisfying multiple constraints. To improve the speed and accuracy of the predictor, a predicted state space is constructed by superposing the offline controllable motion and the online uncontrollable motion, and a space correction is used to eliminate errors caused by the offline data. To improve the smoothness of the thrust and the speed of the searcher, a Global-Local Mixed Search (GLMS) algorithm and a Variable Density Search (VDS) algorithm are designed. Finally, the PSG method is simulated in a landing mission on asteroid 101955 Bennu. Results demonstrate that the proposed method achieves precision landing capabilities under the simultaneous influence of thrust magnitude, direction, and glide-slope constraints, while maintaining computationally efficient guidance command generation.