A Novel LIDAR-driven Two-level Approach for Real-time Unmanned Ground Vehicle Navigation and Map Building

In this paper, a two-level LIDAR-driven hybrid approach is proposed for real-time unmanned ground vehicle navigation and map building. Top level is newly designed enhanced Voronoi Diagram (EVD) method to plan a global trajectory for an unmanned vehicle. Bottom level employs Vector Field Histogram (VFH) algorithm based on the LIDAR sensor information to locally guide the vehicle under complicated workspace, in which it autonomously traverses from one node to another within the planned EDV with obstacle avoidance. To find the least-cost path within the EDV, novel distance and angle based search heuristic algorithms are developed, in which the cost of an edge is the risk of traversing the edge. An EVD is first constructed based on the environment, which is utilized to generate the initial global trajectory with obstacle avoidance. The VFH algorithm is employed to guide the vehicle to follow the path locally. Its effectiveness and efficiency of real-time navigation and map building for unmanned vehicles have been successfully validated by simulation studies and experiments. The proposed approach is successfully experimented on an actual unmanned vehicle to demonstrate the real-time navigation and map building performance of the proposed method. The vehicle appears to follow a very stable path while navigating through various obstacles.