A Road Understanding Algorithm Based on Kernel Region Information and General Knowledge

Aiming at visual navigation of mobile robots in outdoor environment, a road understanding algorithm based on the kernel region information and general knowledge is presented. In this algorithm, kernel region information and general knowledge are used to improve the accuracy of road understanding, and time-concerning impact factor is used to improve the robustness of the algorithm. Based on fuzzy theory, the degrees of membership of road color are allocated to each sub-region, so more precise information can be offered for robot navigation with different degrees of safety. This algorithm is well designed with optimization algorithm, taking the real-timeness of navigation into consideration.