Path Planning Algorithm Based on VSLAM and Target Detection

With the rise of autonomous driving technology, more and more problems need to be studied. After the construction of the map and the identification of obstacles based on visual slam, the path planning based on the map point cloud is the problem that this paper needs to study. In slam, we choose orb-slam3 with IMU added to build a map, and add a target detection module for pedestrian detection. The commonly used algorithms in path planning include A* algorithm, Dijkstra algorithm, DstarLite algorithm derived from A-star algorithm and Weightedastar algorithm. The A* (A-Star) algorithm is the most effective direct search method for solving the shortest path in a static road network. The closer the valuation value is to the actual value, the better the valuation function can be obtained. However, when multiple minima appear, it is not guaranteed to find the optimal solution and will cause a lot of performance consumption. The research direction of this paper is to provide a good algorithm scenario for path planning based on the orb-slam3 map construction and the fusion of target detection modules, and then use the heuristic information to improve the search efficiency through the improved a* algorithm. In this paper, the A* algorithm is used to realize the path planning of the autopilot, and the A-star algorithm is combined with the Stanley trajectory tracking to achieve more efficient path planning.