Formula SAE is an international design and performance competition, organized by the Society of Automotive Engineers. REV converted a petrol-SAE race car fom UWA Motorsport to electric vehicle using a dual rear motor drive with an electronic differential. In a second evolution, REV made this car self-driving. Using an array of LiDARs, Inertial Navigation System and computer vision system, the vehicle is able to drive autonomously around the track or follow a road. Data processing is being conducted with an Nvidia Jetson AGX Xavier as the central node and several auxiliary embedded systems for hardware control.

Our project utilises the Autonomous SAE vehicle as a test bed for technologies including visual SLAM, semantic segmentation, path planning/optimisation and we work continuously to upgrade the capabilities of the car. With the recent advent of low-power GPU accelerated embedded systems, much of our work is focussed on applications of deep learning to real-time sensing and control problems.

Specifications:

Base Vehicle: UWA Motorsport Car Drivetrain: Dual rear motors Controller: Kelly KBL Batteries: 15 ThunderSky Lithium Ion Phosphate LFP90AHA 3.2V Voltage: 48 V Capacity: 4.3 kWh Max current, power: 270A, 13kW Charging Time: 2 hours (full charge) Top Speed: 70 km/h Bat. Management: EV Power BMS Main Processor Nvidia Jetson AGX Xavier (Linux) Framework Hybrid custom C++ / ROS with Hardware-in-the-loop Simulator Embedded Systems Custom drive-by-wire controller, indepedent safety controller (TI Hercules) Visual Sensing Twin FLIR Blackfly global shutter GigE cameras LiDAR Sensing IBEO Lux 4-layer Automotive LiDAR, SICK 2D LiDAR, Velodyne 360-degree LiDAR Localisation Xsens INS w/ integrated GPS fusion, wheel odometry, vision/LiDAR SLAM

Click for larger images.