Research On The Characteristics Of The 3D Feature Extraction Device For Typical Robot Targets

The purpose of this paper is to study the robot typical target three-dimensional feature extraction device in order to provide the robot system with a method to three-dimensional detect the target in the environment.The device combines two-channel ultrasound and binocular stereo vision two detection methods.Among them,the ultrasonic probe judges by the way to the environment of the target distance and orientation,stereo vision in the ultrasonic detection is based on the passive extraction of the target in the environment of three-dimensional information.Ultrasonic measurement has fast speed and stereoscopic vision measurement has high accuracy.This paper combines the advantages of the two above.The use of time-sharing work is to make both of them in the detection function achieve the purpose of complementary.In this way,it ensures the accuracy of three-dimensional feature extraction and makes the system have good real-time.Basing on the analysis of the characteristics of both ultrasonic and stereoscopic vision,the paper designs a dual-processor system with FPGA and ARM as the core of the control of the ultrasonic module and the stereo vision module.The design and the debugging work of hardware and software are completed around the FPGA and ARM.In addition,the thesis also makes a three-dimensional modeling of the structure of the device,the simulation analysis of the dynamic characteristics of the virtual prototype and the finite element analysis of the key components.And the basic parameters of the driving motor are given according to the simulation results.According to the needs of the research,the thesis designs and completes the related experiments.The experimental results show that the ranges of detection of the system and the accuracy of the three-dimensional feature extraction of the typical target satisfy the needs of the mobile robot for the extraction of the target three-dimensional features in the environment.In addition,the device is low in price and it is easy to use,which not only can be applied to the mobile robot platform but also provides referential value for other mobile platforms to detect three-dimensional objects in the environment.