| Summary: | 碩士 === 國立臺灣科技大學 === 機械工程系 === 105 === The control architecture of this visual guided electric vehicle is consisted of a FPGA embedded system for the electric vehicle driving control and a PC for the image processing and data analyzing of laser ranger finder called driving decision
making system. Two systems are communicated with a UART interfere.
In this study, the vehicle visual guided driving process can be classified into three parts. Firstly, the road image ahead of vehicle is captured by using camera. The image pre-processing schemes are used to obtain multi-lane marking image coordinates information. Then, the Homography Matrix technology is used to convert the information into the world coordinate system for target trajectory tracking. Secondly, the vehicle current location is estimated by using the car kinematic model with four-wheels and steering encoders as feedback signals. Then the relative position error between target point and the vehicle current position can be founded. Finally, the pure pursuit geometric tracking control algorithm is used to calculate the steering angle correction command for closed loop trajectory tracking control purpose.
If there is an obstacle in front of the vehicle-driving path, the obstacle avoidance path planning will be generated based on acquired multi-lane image information and the laser ranger finder data. These image processing schemes, lane following algorithm, driving decision making and obstacle avoidance strategy are implemented on a lab made four- wheel drive electric vehicle for campus playground driving test.
|
|---|
