AI Car In Simulation Based On HD-Map

Lane topology information—namely lane connectivity, adjacency, and similar information—is extracted from the HD map to build a topological map for path-finding. The in-memory representation of the navigation map is a directed-graph structure, in which each lane on the road is a node (Node) in the directed graph. If the current lane node has a connection relationship with its predecessor lane, or if crossing is permitted between the current lane node and its left-adjacent lane node (i.e., the boundary line is a dashed yellow line or a dashed white line), or if crossing is permitted between the current lane node and its right-adjacent lane node, then an edge (Edge) is created between the two lanes and added to the directed graph that represents the road topology. Once the topological map is built, paths are computed based on existing path-finding strategies, such as the Dijkstra algorithm, the A* algorithm, and their various variants.