| The acquisition of precise position information in indoor scenes is one of the key challenges of mobile robots.Traditional positioning methods such as GPS,Wi-Fi,Bluetooth,dead reconking,are not ideal for accuracy in indoor robots.However,ultra-wideband wireless positioning system has more applications in indoor positioning due to its high multipath resolution,high positioning accuracy,and low sensitivity to other wireless signals.The main research goal is to improve the positioning accuracy of mobile robots in indoor environments.In order to solve the problem of low indoor positioning accuracy and serious signal interference,a combined positioning of UWB positioning and inertial measurement unit information under the extended Kalman filter framework is proposed based on the advantages and disadvantages of UWB technology and inertial navigation.The state equation and the observation equation are analyzed in detail according to the system model.As of the noise problem of MEMS IMU sensors,a denoising method is propesed using LSTM recurrent neural network.The input gate,forgetting gate and output gate of the LSTM model can save long-term memory in time series and obtain the non-linear relationship in sequence data,which has a significant effect on the noise removal of low-cost IMU sensors.To verify the positioning effect of the UWB/IMU fusion algorithm,a UWB-based indoor mobile robot positioning system was built,which includes the upper computer positioning display platform and design of hardware and software in the robot controling system.In experimental tests,the UWB/IMU fusion solution designed above has a positioning accuracy of 0.12 meters in the line-of-sight environment and a positioning accuracy of 0.43 meters in the non-line-of-sight environment,which improves 42.7% and 47.5% compared with the UWB positioning alone. |
PDF Full Text Request