Research And Implementation Of A Food Delivery Robot Positioning System Based On The Fusion Of Ultra-wideband Positioning And Track Deduction Algorithms

In recent years,as more and more robots have infiltrated into our lives,service robots have been developed rapidly,and the delivery robots have become one of the research hotspots in this field.Due to the complicated construction and inflexible path replacement,the early magnetic navigation delivery robots were gradually eliminated.The non-track navigation technology has become the development trend of the delivery robots.In the non-track navigation technology,the visual navigation has a poor environment adaptability.The cost of laser navigation is too high,and the use of multi-information fusion positioning algorithm has cost-effective advantage,high positioning accuracy advantage,good positioning stability advantages and other advantages.This paper has done a deep research in the UWB positioning technology,track deduction algorithm and multi-sensor information fusion algorithm through the delivery robot.The main research work includes the following aspects:1.Analyze and study the research status of service robots,indoor positioning technologies and multi-sensor fusion algorithms at home and abroad.Analyze their shortcomings and deficiencies,and establish the research direction and structure of this paper.2.Establish the kinematics model and the kinetic model of the delivery robot.And based on the models,deduce the traditional odometry and dead reckoning algorithm for the delivery robot.Improve the algorithm based on the traditional odometry and dead reckoning algorithm in order to make the the improved odometry and dead reckoning algorithm more suitable for the long time and long-distance working conditions of the delivery robot.3.Analyze the existing indoor absolute positioning technology and select the UWB positioning technology based on TDOA algorithm which is suitable for the restaurant environment;In order to improve positioning accuracy and stability effectively,this paper fuse the positioning results of the UWB system with the positioning results of the improved odometry and dead reckoning algorithm through the extended Kalman filter fusion algorithm.4.On the Windows side,the TDOA algorithm is used to calculate the UWB data and get the positioning results.At the same time,the delivery robot is controlled by ROS on the Linux side;The hardware circuit and the bottom driver of the maincontrol board are designed so that it can receive the left and right wheel speeds send from ROS platform and the improved odometry and dead reckoning algorithm is used to obtain the positioning results of the delivery robot.5.In the experiment,the TDOA algorithm is used to calculate the UWB data and obtain one set of the positioning results of the delivery robot.At the same time,the other set of positioning results of the delivery robot were calculated by the improved odometry and dead reckoning algorithm.Analyzed their errors and fused them through the extended Kalman filter fusion algorithm,and the fused positioning result's accuracy and stability are greatly improved,its positioning error is about 15 centimeters which is far less than the positioning error of the UWB positioning system and the improved odometry and dead reckoning algorithm.