移动机器人多传感器数据融合方法研究OA北大核心CSTPCD

Multi-sensor data fusion technology can better solve the problem of data incompatibility generated by smart devices,im-prove device operational efficiency and productivity.Existing methods lack high-performance performance and do not consider data privacy protection issues.It innovatively introduced federated learning into multi-sensor data fusion.The federated learning local model used the Gated Recurrent Unit(GRU)algorithm to solve the multi-sensor data fitting problem.A novel parallel stere-oscopic multi-sensor data fusion method was designed for the first time,which has excellent fusion performance and ensures the privacy of each client's data.The experimental results demonstrate the correctness and rationality of this method,as well as its ad-vantages in robustness.