Font Size: a A A

The Design And Implementation Of RoboCup Humanoid Robot Upper Control System

Posted on:2021-01-23Degree:MasterType:Thesis
Country:ChinaCandidate:H ShiFull Text:PDF
GTID:2518306476952879Subject:Control Engineering
Abstract/Summary:PDF Full Text Request
Humanoid robot is similar to human in terms of body structure,perception,and action,and their research field involves many academic fields such as machinery,control,computer,which is worthy of in-depth research.The Robo Cup is designed to beat the human World Cup champion,and the project has given a boost to robotics,particularly humanoid robotics.Based on SEU_Uni Robot humanoid soccer robot,the upper control system applied to Robo Cup humanoid Kid Size group was optimized and realized.On the basis of presenting an overall scheme of the upper control system,the design of each functional module in the system is discussed in detail,and the design of the simulation platform for experimental debugging is introduced.According to the competition environment and rules,the requirements of the upper control system are analyzed to make sure that the robot needs to be able to sense the environment,position itself and make decisions.Therefore,the upper control system is composed of visual module,positioning module and decision module.The vision module processes the acquired image to obtain the information of observations required by the positioning module.The object in the image is acquired by extracting the spare box,classifying the spare box and integrating the extraction box.The object in the image is converted into the information of observations through the monocular camera ranging.The positioning module is divided into two steps: preliminary estimation and deviation correction,and the error correction link is designed from two directions particle filter and kalman filter.The decision module is the nerve center of the robot,which is divided into three levels team decision,individual decision and execution planning.The method of reinforcement learning is used to complete the team decision planning,the finite state machine is used to realize the individual decision,and the Markov decision process is used to realize the execution planning.A simulation platform is built for the experimental debugging of upper control system,which has the functions of noise simulation,data recording and analysis.In addition,external Webots platform interface is provided to expand the experimental environment from 2D environment to 3D environment.
Keywords/Search Tags:humanoid soccer robot, machine vision, filtering positioning, reinforcement learning
PDF Full Text Request
Related items