Research and Simulation of Millimeter Wave Radar Localization Method for Indoor Mobile Robots

With the rapid development of robotics and sensor technology, indoor mobile robots have become important in the field of robotics, so it is especially important to study the positioning methods of mobile robots. In this paper, we first analyze the bottlenecks of indoor mobile robots and the localization problems that need to be solved; then, we design an online low-complexity algorithm for super-resolution localization using one frame of IF signal to address the shortcomings of localization technology, and verify the performance of the algorithm through numerical experiments. The relative motion trajectory of the object is calculated for one frame of data provided. The orthogonal matching pursuit algorithm requires a pseudo-inverse of the matrix, and the dictionary needs to be very large to ensure the reconstruction accuracy, resulting in a high complexity of matrix operations. To reduce the complexity, the most direct way is to reduce the size of the dictionary matrix, which can use the information of the previous frame to estimate the distance and angle range of the possible target in the next frame. Finally, the discrete data points are then linked together in a data association manner. Simulation results show that the proposed method has significant improvement in the probability of target detection as well as localization accuracy.