The Autonomous System (AS) sub-team is responsible for all aspects that make our vehicle driverless. The AS members develop the software and set the hardware (i.e. sensors, actuators etc.) that in combination enable our race car to operate autonomously – without a driver. Through a set of sensors and cameras, the vehicle gathers location data from its external environment, interprets them using machine learning algorithms, plans its path accordingly and actuates the pedals and the steering wheel through dynamic control. Specific areas of interest of this sub-team include path planning, machine learning and machine vision, odometry, localization among others.
Our goal in the autonomous system is to present a competitive, fully autonomous racing vehicle that can complete the competition’s courses as fast as possible. To achieve this, we heavily modify aristurtle’s previously manufactured electric vehicles. We develop a drive-by-wire system that actuates the vehicle without mechanical support from a human and employ a powerful, on-board processing unit capable of real-time computer vision and optimization. We also equip the vehicle with environment sensors, such as lidar and cameras to perceive the track boundaries and vehicle state measuring sensors, such as IMU, GNSS and wheel rotation speed sensors. This data is fed to the computer and, through careful preprocessing, the estimation pipeline. After a map of the perceived environment is created, the control algorithm computes an optimal driving policy and sends signals to the low-level controllers of the actuators to perform the desired actions. A simulated environment is developed to test algorithm configurations more efficiently.