Research On Adaptive Vehicle Positioning Based On Double Odometers

The vigorous development of information and communication technologies such as 5G and cloud computing will detonate the transformation of people's travel modes.Smart cities and intelligent transportation have become the main development trends in the future.The performance of the vehicle positioning and navigation system affects the traffic safety and communication efficiency of intelligent transportation directly.Therefore,the public has a more urgent need for highquality vehicle positioning services.From the aspect of economy and practicality,this paper chooses low-cost odometers and GPS receivers for vehicle navigation,and combines the Cubature Kalman Filter algorithm with the Interacting Multiple Model algorithm to construct the double odometers / GPS integrated navigation system with adaptive functions.The main tasks are as follows:(1)Investigating the geometric position relationship of the double odometers fixed on the vehicle-driving wheels relative to the vehicle,combining this geometric constraint with the odometer dead reckoning,and further observeing the posture of the vehicle to reduce the observation error.In addition,the errors of the double odometers and GPS receivers are analyzed respectively.Based on this,the double odometers / GPS integrated navigation system is researched and designed.The GPS is used to complete the initial alignment and error correction of the double odometers inertial navigation.This paper combines the integrated navigation system with the Cubature Kalman Filter algorithm and uses MATLAB to perform simulation.The simulations show that compared to the single GPS observation,the positioning accuracy of the integrated navigation system has been improved greatly,and the system's estimated time delay of the speed information also reduced by about 5s.(2)The actual maneuvering state of the vehicle is difficult to describe with a single motion model.This paper further improves the double odometers / GPS integrated navigation system and combines it with an Interacting Multiple Model algorithm.The original observation data of the double odometers and the filtering residual of the Cubature Kalman Filter algorithm are used to judge and modify the vehicle motion model's probability.The simulations show that compared with the ordinary Interacting Multiple Model algorithm using the Markov chain to control model probability method,the improved algorithm has improved the judgment efficiency of the model greatly,and the positioning accuracy has increased by about 36%.(3)Considering that there may be abnormal situations of the observation source during the driving,this paper designs corresponding alternatives for different types of observation information loss.This design structure has a strong adaptative ability,which can ensure that the observation information of the system is always output continuosly.Simulations show that when the information of the observation source is incomplete,the positioning error of the system increases slightly,meanwhile it still provides a good positioning estimate as a whole.This paper can provide a reference for improving the accuracy of vehicle positioning,and also provide a reference for obtaining accurate and real-time vehicle posture information in intelligent transportation systems.