Multi-robot Fusion with Measurements Compensation Based on Recursive Least Square

Communication network for multisensor system often produces missing data in the process of transmitting local information from the local sensor to the central processor because of it intrinsic property. The traditional methods to deal with missing data are to abandon this information in the fusion process, so its fusion accuracy is reduced. Aiming at this problem, by introducing the batch and recursive least square (LS) two fusion algorithms are proposed to improve the accuracy of the fusion estimate. Because the first proposed method based on the batch LS performs fitting at the same time, so its computational performance is bad. And due to the second method based on the recursive LS fitting possesses good computational property one by one, so it has high real-time performance and fast running ability. The analysis and simulation for three algorithms show that the proposed two algorithms to deal with the case of missing data is valid and they has the same fusion estimate accuracy.