Complex terrain perception based on Hidden Markov Model

Terrain perception in complex environment is important for Autonomous Land Vehicle to drive automatically. In order to access the terrain information, in this paper, we present a terrain perception method based on Hidden Markov Model (HMM) which combines LIDAR with machine vision. On the basis of spatial fan-shaped model, terrain feature extraction is performed to acquire the observation model. Hidden markov models describe the vertical structure of the driving space and Viterbi algorithm is used for terrain classification. Then the navigation decision is given based on the perception of the complex environment. Experiment results show that the method can give an accurate environment description for ALV.