| This thesis uses FinaRobot,the financial robot of the University of Electronic Science and Technology of China,as an R&D platform.FinaRobot assists Finance Department staff to provide teachers and students with online consultation on financial issues through anthropomorphic human-computer interaction to ensure that their autonomous service is not affected by the automatic shutdown caused by low battery.This article has designed a set of automatic charging system for it to complete the remote navigation and short-range charge docking two parts of the work to achieve automatic charging.In addition,according to the user's location guidance requirements,a set of intelligent navigation system that integrates voice interaction is designed.In this system,the robot needs to use voice to communicate with users,and at the same time,it can plan a reasonable navigation path and achieve obstacle avoidance.Therefore,this article will focus on research and experimentation on the construction,positioning and navigation of mobile robots.The detailed work and contributions completed in this thesis are described as follows:(1)Concerning the functional requirements of the navigation system in this thesis,the overall architecture(including the software system and the hardware system)of the financial robot FinaRobot is conceived and designed.The most important software system mainly includes three modules: Environment map creation,voice navigation With automatic charging,Socket communication is selected as the network communication method in the two navigation systems.(2)Introduced the basic principles of SLAM and Lidar.The traditional RBPFSLAM has the problems of high computational complexity and particle degradation.An improved RBPF-SLAM algorithm is proposed,which uses an improved particle resampling strategy to reduce the speed of particle degradation,maintain particle diversity,reduce the number of particles,and further reduce computational complexity.Based on the above improved algorithm,code writing,and finally select two different environments to verify the optimization effect of the algorithm.Finally,the workflow of how to create an environment map under ROS is described in detail.(3)It proposes a navigation method that integrates speech interaction and natural language processing.After obtaining the user's speech navigation requirements,it is converted into textual information through speech recognition technology,and then the environmental map of the annotation destination point is fed back to the natural language processing.The user finally enters the navigation mode.After guiding the user to the destination,feedback information is given to the user through speech synthesis technology.Finally,the design and implementation of a smart navigation system based on voice interaction was completed.(4)A charging pile alignment algorithm based on mutation point detection was proposed.It can rely on existing lidar sensors to achieve higher accuracy of charging pile alignment.At the same time,a set of charging heads and charging bases with a relatively high tolerance for error was designed.Fully increase the success rate of accurate docking for automatic robot charging.At the same time,the design and implementation of an automatic charging system have been completed.(5)Finally,build and integrate the navigation system software module on the existing hardware platform,and made a series of test experiments for the complete robot navigation system to verify the correctness and effectiveness of each software module. |
PDF Full Text Request