A novel space target-tracking method based on generalized Gaussian distribution for on-orbit maintenance robot in Tiangong-2 space laboratory

Abstract

Accurate target tracking based on visual images is the key for intelligent robots to assist or replace astronauts to work in space station. However, the special space environment such as non-uniform illumination and high-energy particle radiation is a huge challenge, which may lead to complex noise coupling in vision image. This paper proposes a novel method for accurate target tracking, the essence of which is the Retinex image enhancement algorithm in CIELAB color space (LAB-GRetinex) and the generalized maximum correntropy Kalman filter (GMCKF) which are all based on generalized Gaussian distribution. The LAB-GRetinex algorithm chooses the CIELAB color space, which is closer to the human vision, as the processing color space, and the generalized Gaussian distribution can estimate the light image accurately, so the influence of non-uniform illumination can be reduced effectively. Meanwhile, the GMCKF algorithm adopts the generalized correntropy criterion based on the generalized Gaussian distribution to replace the minimum mean square error (MMSE) criterion to realize the optimal filtering effect under the complex non-Gaussian noise, which can improve the target tracking accuracy. Sufficient ground simulation experiments and application experiments in the Tiangong-2 space laboratory verify the effectiveness of the proposed algorithm which can track the target accurately in the special space environment and provide the precise pose information for on-orbit robot maintenance verification. This research lays a technical foundation for the application of intelligent robot in the construction and operation on space station in the future.