Collision Avoidance using Lane Departure Warning and Obstacle Determination in the Automotive Environment

Statistics show that worldwide motor vehicle collisions lead to significant death and disability as well as significant financial costs to both society and the individuals involved. This thesis aims to address this issue by proposing a system for the detection of lane departures and obstacles, two of the leading causes of road fatalities by collisions. Awareness of the environment that surrounds a vehicle can make driving and manoeuvring of the vehicle much safer for all road users. The primary focus of this thesis is the development of an image processing system to detect lane departure using a single, forward facing visible spectrum camera. This system uses a novel lane marking segmentation algorithm in accordance with national standards for lane markings and identifies lane departure using guidelines issued by the International Organisation for Standardization. Issues with a single camera-based LDW system can arise when other road users occlude the field of view of the sensor, or by glare from oncoming headlamps and bright sunlight on the sensor. To address these problems, a lane departure warning system that utilises multiple cameras is also presented. This multi-camera system is more robust to errors caused by lane marking occlusions, sensor failure, and glare that a single camera-based system can suffer from. Object detection is a large area of research in the field of automotive technologies. By displaying the surroundings of the vehicle and the included objects to the driver in a useful and intuitive way, this aids the driver to safely manoeuvre the vehicle. Similarly, warning the driver of potential obstacles in the trajectory of the vehicle reduces the probability that the vehicle will be in a collision. An examination of the state of the art in object detection and distance determination technologies is presented and from analysis of the limitations in existing systems, a number of novel obstacle detection/distance determination systems are also proposed in this thesis. Overall, this research aims to reduce the number of vehicle collisions and contribute to a safer driving environment