2019 EV 'Chicken' turned autonomous

Development car | No competitions open yet

2021 AV Rooster was built off the back of our succesful 2019 EV, 'Chicken'
‍
UQ Racing has been in the process of developing an autonomous vehicle (AV) for the Formula Student competition. Since 2020, we have been making great progress in the fields of simulation, actuation and sensor processing. Our team's goal for 2022 is to lay the technical foundations for our autonomous vehicle, providing the groundwork for future developments involving more advanced perception and path planning.

Currently our team has a fully functional simulation of the AV, including its vehicle dynamics and an emulation of the perception and actuation pipelines of the car. Both our simulation of the vehicle and its on-board software are written in C++ and run on ROS Noetic. We use the simulation to design and develop new algorithms and validate the performance of existing ones. For example, we have developed and verified a path planning algorithm in the simulation that uses mathematical optimization to calculate a fast route around the track for the autocross event.

Our car has a variety of sensors that we have been actively working on to process effectively. We have a LeiShen CH128X1 hybrid solid-state LiDAR to perceive and map the track in 3D. We also have an Intel RealSense depth and RGB camera for vision-based cone detection. There is also an SBG Systems Ellipse-N Inertial Navigation System to determine the car’s position on the track accurately for navigation. Finally, we have a Neousys Nuvo-7162GC ruggedised PC for processing all the car’s data at real-time speeds.

We have been developing and refining our AV’s actuation system through an extensive process of design and review. The two main systems in development of our actuation system are the steering actuation and the emergency brake system (EBS). The steering actuation uses a Moog SmartMotor and the EBS uses a pneumatic system. To ensure the safety and reliability of this vehicle, we are also developing our own Autonomous Fault Management Module, which contains programmable and non-programmable logic to ensure the vehicle’s subsystems are always fully operational.

Once the actuation and safety groundwork for the AV has been developed effectively, we aim to improve the vehicle’s software to create an effective and competitive entry into FSAE. Our next projects are mainly improving our visual cone detection algorithm, which will in turn allow us to map out the track when paired with our INS. We are also going to be focusing on improving our path planner to be flexible across multiple autonomous missions.