High-precision position predictive control of mobile robot based on LSTM algorithm

In this paper, a high precision position predictive control of mobile robot is proposed based on LSTM (Long short term memory) deep learning neural network algorithm. The structure-based error back-propagation algorithm and iterative learning are introduced to make the actual predicted position of mobile robot better approximate the expected position, and obtain satisfied performances. Due to the unique ability of LSTM algorithm to memorize the information far away from the current time, the current position can be better predicted by processing the trajectory information of the mobile robot which is far away from the current time, and make the predicted value closer to the real value. The simulation demonstrates that the proposed method has the excellent predictive control performance.