Implementation of the autonomous functionalities on an electric vehicle platform for research and education

Self-driving cars have recently captured the attention of researchers and car manufacturing markets. Depending upon the level of autonomy, the cars are made capable of traversing from one point to another autonomously. In order to achieve this, sophisticated sensors need to be utilized. A complex set of algorithms is required to use the sensors data in order to navigate the vehicle along the desired trajectory. Polaris is an electric vehicle platform provided for research and education purposes at Aalto University. The primary focus of the thesis was to utilize all the sensors provided in Polaris to their full potential. So that, essential data from each sensor is made available to be further utilized either by a specific automation algorithm or by some mapping routine. For any autonomous robotic system, the first step towards automation is localization. That is to determine the current position of the robot in a given environment. Different sensors mounted over the platform provide such measurements in different frames of reference. The thesis utilizes the GPS based localization solution combined with the LiDAR data and wheel odometry to perform autonomous tasks. Robot Operating System is used as the software development tool in thesis work. Autonomous tasks include the determination of the global as well as the local trajectories. The endpoints of the global trajectories are dictated by the set of predefined GPS waypoints. This is called target-point navigation. A path needs to be planned that avoids all the obstacles. Based on the planned path, a set of velocity commands are issued by the embedded controller. The velocity commands are then fed to the actuators to move the vehicle along the planned trajectory