High-Precision Autonomous Navigation Method for Deep Space Probe Cruise Phase Under Uncertain Conditions with Q-Learning Filter
摘要
The cruise phase of deep-space exploration missions will play a vital role in future planet exploration tasks. The navigation system model is constructed based on the starlight angular distance and line-of-sight vector measurement models. To overcome the uncertainties associated with measurement noise in the cruise phase, a Q-learning unscented Kalman filter is applied to improve the autonomous navigation accuracy through the optimal selection of the noise parameter model. Simulations of a Jupiter exploration scenario demonstrate the performance enhancement of the proposed navigation scheme.