||In this study, the autopilot system of the car is realized by the use of a personal mobile computer and embedded development platforms, such as 86Duino-One and TMS320F2812. In lateral control, lane keeping could be realized by the path and speed planning system using differential GPS, which computes lateral offset and head angle error information sent to lateral fuzzy controller. Lateral fuzzy controller computes desired steering wheel angel to let the car track the reference route perfectly. |
In longitudinal control, speed and acceleration of the car can be obtained from a encoder. Difference between planning speed and instant speed is one of the inputs of longitudinal fuzzy controller, the other is the difference between planning acceleration and instant acceleration. Then, longitudinal fuzzy controller computes the desired throttle value to let the car track the planning speed.
In path and speed planning system, the road in NSYSU campus is the reference route. There are slight bend, sharp bend, and straight road in the reference route. Comparing to the route, the curvature of the present car position can be calculated. Then, the system can adjust planning speed in real time in accordance with variety of curvature. When an obstacle appears in front of the car, the system would get the information of relative distance with 2D-lidar. Relative distance and curvature of the reference route for path and speed planning system make the system adjust appropriate self-driving speed in real time. If relative distance is less than the safety distance 5 meter, the system would stop the car to reduce damage of impact. The experimental results show that the proposed system can successfully achieve the above functions.