Path following and obstacle avoidance for autonomous vehicle based on gnss localization

Abstract

Nowadays, many systems, formerly operated by human beings, are now developed to minimize user control effort. One outstanding system that receives close attention is the autonomous driving system. However, several autonomous driving systems that are currently developed utilize expensive sensing devices, e.g., camera and laser scanner. Moreover, these devices cannot be solely employed but a high-performance processing unit is also required. These pricey components result in an expensive system that does not worth to be used in some practical applications. Therefore, this research intends to utilize other low-cost sensing devices so that the final price of the developed system can be reduced. Hence, the Global Navigation Satellite System (GNSS) with a Real-Time Kinematic (RTK) correction was applied to this research as a prior sensor. However, a laser scanner was still employed as a complementary sensor to detect obstacles which cannot be detected by the GNSS alone. This developed system was designed to effectively operate at a travel speed lower than 15 kilometers per hour in a GNSS-friendly environment. In this research, the micro electric vehicle was modified by installing the steering control system and the speed control system, which consists of the acceleration control system and the braking control system. These supplementary systems are controlled by the high-level control system. Next, the high-level control system software was developed. This software controls a vehicle to follow a predefined route by using the GNSS in a localization process and using a laser scanner in the obstacle avoidance algorithm which was developed in this research, i.e., the Scored Predicted Trajectory. Then, the parameters, which affect the high and low-level control system characteristic, was tuned until a satisfactory response was achieved. Next, the developed autonomous navigation system evaluation experiment was conducted in a controlled environment area by separately evaluated the path following system and the obstacle avoidance system. After the path following system experiment was launched using a travel speed of 10 and 15 kilometers per hour, it has been found that the developed system effectively performs even some portions of the test track are either covered by large trees or surrounded by buildings, i.e., the environment by which the performance of the GNSS is degraded. The result of this experiment shows the average deviation distance from the waypoint of about 10 centimeters. In the obstacle avoidance system experiment, the result shows that the developed system responds to the obstacle by evading it and safely converging to the predefined path according to the designed algorithm.