Control Aware Path Planning For Self-Driving Car

Abstract

Self-driving cars represent a significant advancement in autonomous systems, aiming to revolutionize transportation by enhancing passenger safety and reducing human-driving errors. While existing research has focused on trajectory tracking control using pre-defined paths, this thesis extends the capabilities of self-driving cars by introducing a novel approach to controlaware path planning. The primary objective of this research is to enable self-driving cars to autonomously plan their paths without tracking pre-defined trajectories. Instead, the vehicles will embark on a journey at a user-set initial speed while continuously assessing their surroundings. The critical innovation lies in the development of a real-time path planning algorithm that can dynamically adjust the car's trajectory based on information obtained from the car's sensors and previous knowledge from the vehicle's own behavior (plant). To achieve this goal, the thesis proposes a control-aware path planning strategy that factors in the vehicle's current state, environment perception, and knowledge derived from past experiences. The path planner employs advanced path planning algorithms and strategies that are based on latest researches to anticipate and respond to sudden obstacles and different scenarios in the car's path, such as vehicles braking abruptly or unexpected road conditions. When the system detects a potential obstacle ahead, it computes a new path that allows the car to avoid any kind of obstacle like car, traffic cones, or humans while adhering to safety constraints and user preferences. Crucially, the planning algorithm optimizes for reaching the new path while taking into account the vehicle's current state, speed, position for real-time implementation. To validate the proposed control aware path planning approach, comprehensive simulation studies are conducted using highfidelity driving simulators CARSIM and Simulink. The results demonstrate the effectiveness of the algorithm in enabling self-driving cars to adjust their speeds and paths, ensuring smooth navigation and collision avoidance even in complex and challenging environments. The contributions of this master's thesis offer a significant step forward in self-driving technology. By improving self-driving cars with control-aware path planning capabilities, the research enhances passenger safety, provides a more reliable and adaptable driving experience, and opens the door for future developments in the field of autonomous vehicles.