Design And Analysis The Algorithms Of Smart Car In Environmental Perception And Interaction

More and more people believe that automatic driving will be the technology which can change the world nowadays.As a basic component of automatic driving technology,people study environmental perception with great passionate.Automatic driving technology has many parts,including traffic sign recognition,lane line and pedestrian detection,interaction between cars,security assurance and so on.We study the formation control,traffic sign recognition and strategy of environmental perception.The main work and results are introduced as follows:1.Chapter 2 talks about formation control.We analysis the strategy of formation control and choose the leader-follower strategy.Then control algorithms of smart cars,including input-output feedback linearization and PID,are discussed.We focus on the problems of joining the formation and leaving the formation.The framework we proposed is named CJL,we do some simulation to valid it.2.Chapter 3 studies the topic of traffic sign recognition.Inspired by the rapid development of deep learning,we want to apply it to the portable device.Traffic sign recognition is the theme of this part.Different architectures of the network are designed.Experiments platform is raspberry pi 3.We test some of the classifiers.Finally,we design a network named TSNet and prove-its efficiency for portable device.3.Research about the strategy of environmental perception is raised in chapter 4.We believe that the perception ability of smart cars is important,and the devices on road need this ability too.Then the devices on the road can send the information of the environment to cars.It can improve the stability of smart cars.When some abilities of the smart car are lost,automatic driving can still be kept.This will be important for some special circumstance.