UWB Indoor Positioning of Mobile Robot Based on GA-BP Neural Network

BP neural network has good performance in robot UWB positioning, but it is easy to get into local extreme value. To solve the problem, the weights and threshold randomly generated by BP neural network are optimized with genetic algorithm. With the BP neural network and the optimized GA-BP neural network for mobile robot localization experiment, the optimized GA-BP neural network can overcome the defects of BP neural network. In indoor line-of-sight (LOS) and non-line-of-sight (NLOS) environment, the mean positioning errors of the optimize method were reduced by 46% and 24% respectively; Under the same probability condition, the positioning error of GA-BP neural network in LOS environment is about 48% lower than that of BP neural network, and in NLOS environment is about 20%.