| dc.description.abstract | Autonomous driving requires perception and control. The first part of the dissertation is focused on an aspect of control related to automatic vehicle following that is not well understood, namely, the influence of imperfect wireless connectivity in vehicle platooning applications. The primary goal of most research in vehicle platooning is to enable the shortest inter vehicular spacing while maintaining safety, since short following distances are known to improve fuel efficiency and traffic mobility. It is also known that wireless connectivity can be exploited to achieve tighter platoon formations, but the effect of imperfections of wireless links on platoon stability were not well understood. This work proposes an algorithm to estimate the smallest time headway that guarantees safety based on the average packet reception rate. The algorithm has been corroborated using Model in Loop (MIL) simulations as well as test runs with a hybrid car. This thesis also develops a method to estimate the maximum perturbation in spacing error of any vehicle in a string stable platoon based on the lead vehicle's acceleration maneuver. This allows a designer to pick a safe standstill distance.
The second part of this dissertation explores the challenges of environment perception and sensor fusion under adverse visibility conditions for autonomous driving. The sensor stack for autonomous vehicles usually consists of on one or more of radars, visible spectrum/RGB (Red-Green-Blue) cameras and lidars. RGB cameras perform poorly in low light conditions (at night) as well as in direct sunlight. While automotive radars are resilient to environmental conditions, they only offer a low resolution output. In this thesis, we explore the benefits of combining a Long Wavelength Infrared (LWIR) thermal camera with a radar sensor for detection and tracking of vehicles/pedestrians in poor visibility conditions. A modified Joint Probabilistic Data Association (JPDA) filter is implemented on real-world data to demonstrate the feasibility of the proposed system. | en |
