An Optimization-Based Robot Navigation System in Complex Terrain on a Hybrid Map

In the field of robot navigation, grid-based maps are primarily leveraged in systems operating in complex terrain. However, the efficiency of the path search module in these maps is generally low due to the numerous conditions, e.g., climbing and tilting angles, that need to be checked among grids, hindering the real-time performance and stability. In addition, the discontinuity of grid-based maps also impedes the implementation of the path optimization module, significantly reducing the quality of the optimized path. Although the curved surface can be used to relieve the aforementioned two issues, the actual terrain surface is variable and cannot be represented by a binary function. Furthermore, the presence of vertical obstacles can intensify these challenges. To this end, we propose a stable robot navigation system in complex terrain on a hybrid map (HMRNS-CT). A hybrid map, incorporating a grid map, is utilized to extract vertical obstacles through variance analysis. And an adaptive curved surface quadtree map is further employed to fit the rest area. Therefore, in the path search module, the climbing and tilting angle can be checked once only on the fitted curved surface. And a proposed path optimization algorithm can also be adopted to further improve the smoothness and reduce the height change of the path by minimizing a cost function rather than finding the shortest path. A series of simulations and experiments in real-world environments are conducted to evaluate the efficacy of HMRNS-CT.