Dead Reckoning Algorithm Based On The Wheel Mounted MEMS IMU

Low-cost,high-precision and highly reliable autonomous navigation system is crucial for efficient operation and large-scale popularization of mobile robots.MicroElectro-Mechanical system Inertial Measurement Unit(MEMS IMU)lies at the core of a multi-sensor fusion navigation framework due to its low cost,small volume and ability to percept three-dimensional motion of vehicle continuously and stably.However,the navigation error of conventional Inertial Navigation System(INS)accumulate rapidly because of the significant noise and bias instability of the MEMS IMU,which leads to restricted development of its application.Inspired by rotation modulation technology,a Wheel mounted IMU(WheelIMU)-based Integrated Navigation System(Wheel-INS)is designed and implemented in this work,which can improve the independent dead reckoning ability of MEMS IMU thereby let IMU play a bigger role in wheeled robot navigation.The main research work and contributions of this paper is as follows.1.To solve the problem of the installation error of the Wheel-IMU,a mounting angle(the difference between IMU coordinate system and wheel coordinate system)offline calibration algorithm and a lever-arm(the misalignment between wheel center and IMU center)online calibration algorithm are designed and implemented in this paper,respectively.Experiments show that the IMU mounting angle error can be effectively calibrated by the offline calibration approach,and the calibration results have high repeatability.The lever-arm error can be reliably and stably estimated in real time,whose internal accord accuracy is at submillimeter level,which provides the prerequisite for the Wheel-INS to play its role effectively.2.Taking full advantage of the motion characteristics of the Wheel-IMU,there different constraint algorithms are designed in this work,including wheel center speed constraint,displacement constraint and wheel touch point zero-velocity constraint.Detailed derivation process is presented and the equivalence of the three algorithms is also analyzed from the perspective of observation information.Experimental results show that the relative positioning accuracy of the three algorithms is similar,whose maximum horizontal position drifts are all between 1% and 1.5% of the total moving distance.Compared with the traditional odometer aided inertial navigation system in which the IMU is placed on the vehicle body(ODO/Body-INS),the Wheel-INS is less sensitive to the IMU constant bias.Especially,the Wheel-INS can show better positioning performance when the wheel structure is more stable and the rotation speed is higher.3.To solve the problem that the Wheel-INS is not stable enough and only applicable to scenes that the vehicle moves on the horizontal plane,a multi-system integrated dead reckoning algorithm based on vehicle mounted multiple MEMS IMUs(including double Wheel-INS,Body-INS + single Wheel-INS and Body-INS + double Wheel-INS)is designed and implemented to realize information complementation and mutual constraints among multiple systems.The high feasibility of the proposed scheme is proved by experimental results,and it also illustrates that the multi-systems constraints can provide the weighted average results of subsystems,which can show more stable positioning performance.In the multi-systems which contain traditional Body-INS,Wheel-INS can utilize the horizontal attitude information of vehicle provided by the Body-INS to expand positioning ability from two dimension to three dimension,so that the Wheel-INS can get rid of the limit of horizontal motion.A comprehensive and thorough research on the wheel mounted MEMS IMU based dead reckoning system is presented in this paper,including design and implementation,as well as detailed comparison analysis of different algorithms and different IMU configurations,which provides technical preparation for the application of Wheel-INS in the navigation and positioning of wheel-based vehicles.