Research On Vehicle Vision Navigation And Control System Based On Least Square Method And Support Vector Machine

When traditional logical business can no longer meet the needs of enterprises,people will try to find more complex,adaptable and large-scale algorithms to solve the needs of production and life.In the wave of artificial intelligence development in the 21^st century,machine learning,especially deep learning,is widely used in the unmanned driving,involving computer vision,communication,coordination and control.The current unmanned vehicle algorithm requires high-performance hardware,high development cost and great security risks in the research process.To this end,our paper simplifies the research model and focuses on the research of key technologies.From multiple perspectives,combined with navigation algorithms,high-precision obstacle avoidance algorithms and flexible control algorithms,we have implemented a visual-based navigation and adaptive control system.Main tasks are as follows:?1?A new framework for building-climbing visual navigation?BCVN?is proposed.In the building process,the goal is to find the base edge with high confidence.In the climbing process,the goal is to resist environmental disturbances,to extend new smooth edge points along the base edge to calculate navigation line location.In addition,the calculation of deviation and curvature is designed.The deviation is used to control the steering the wheel,and the curvature is used to adjust the speed.For the irregular shaped obstacles on the road,a combination of CNN network and SVM is designed to achieve high-precision obstacle avoidance.?2?In the aspect of steering control,the traditional PD algorithm is improved,and by introducing the segment P and the fuzzy differential D,it is possible to achieve early prediction in steering.This also ensures that the front end has a small plunging angle at the corner,achieving the shortest path of the cornering and reducing the jitter.For the speed control,the traditional PI algorithm is improved,and the compensation control is added to realize the rapid growth and stability.?3?Completing the hardware construction of the vehicle,the design of image debugging software and vehicle embedded platform software.Testing the hardware and software performance of each part of each part of the vehicle intelligent navigation and control system.Tests have shown that the system is in good operating condition and the vehicle is stable on the road.All aspects of the function have met the design requirements.