Markov chain-guided ray explorer algorithm
Hongjun Wang, Xujun Guan, Chuangbo Hao · 2025
Aiming at the trade-off between modeling accuracy and computational efficiency in lidar multi-sensor fusion positioning for target areas, this paper proposes the Markov Chain-Guided Ray Explorer Algorithm (MC-RaE). Through cross-domain technology fusion of probabilistic search, geometric tracking, and data dimension reduction, the algorithm enables efficient modeling of the target positioning area under ranging and angle measurement errors from multiple lidars. Simulation results show that the total computation time of the MC-RaE algorithm is within the millisecond range. In scenarios with lidar ranging errors or angle measurement errors, the volume of the PCA-driven rotating bounding box constructed by MC-RaE is reduced by an average of approximately 10% compared to traditional axis-aligned bounding boxes. This approach provides an accurate and efficient target area modeling solution for lidar multi-sensor fusion positioning.