Advanced Methodology for Tactical Level Decision Making In Autonomous Ground Vehicles

Abstract

The field of intelligent transport system is quite vast and it covers a lot of different sub field including, vision systems, communication systems, trajectory generations system and many more. Out of this vast range, this thesis presents a methodology to devise a single acceleration command based, on-line trajectory planning methodology for overtaking maneuvers of autonomous vehicles ensuring, accident avoidance, passenger comfort, emergency response, and time reduction. The focus is on two autonomy aspects of overtaking maneuver: (i) overtaking in the possible presence of any other vehicle/vehicles in the passing lane, while ensuring passenger comfort, and (ii) ensuring that a safety distance is always maintained between vehicles in order to avoid any collision. The proposed scheme is applicable in real-time, uses environment inputs for guidance and ensures accident avoidance. It also guarantees that passenger comfort remains in safety bounds. Modified Rendezvous Guidance technique is used as guiding principle, the passing/overtaking vehicle is guided in real-time to match the position and velocity (i.e., rendezvous with) of a virtual target during the overtaking maneuver. The virtual target’s position and velocity are determined based on real-time information from the sensors. The sensors gather information about the slower vehicle ahead of the pursuer vehicle as well as other vehicles which may be travelling in the overtaking lane. Collision avoidance is also carried out using the same guidance principle. In fact, a single acceleration command is used for both obstacle avoidance and trajectory generation in contrast to some other trajectory generation methods. The proposed method can be used as a fully autonomous system or simply as a driver-assistance tool. The proposed method was implemented in both computer simulations and experiments. The results clearly demonstrate the tangible effectiveness of the suggested method.