Team Ampera pushed through various prototypes before creating a Raspberry Pi-powered car capable of navigating the Formula SAE Brazil track completely autonomously.
Ampera, a Formula Student team from the Federal University of Santa Catarina, Brazil, has been competing in the electric vehicle (EV) category of the Formula SAE Brazil Competition since 2012. But in 2020, they were inspired by European competitions to start a research group focused on driverless cars. They began developing algorithms for autonomous cars in simulations, and by the end of 2022 they were able to begin implementing the system in their AMP-222 prototype.
The team’s timing proved perfect: in 2023, SAE Brazil decided to include an autonomous showcase in the competition. Ampera was the only team that successfully completed it, entering their new prototype, the AMP-223. In fact, Ampera’s Raspberry Pi-controlled car was the first autonomous vehicle ever to compete in the Formula SAE competition.
How does an autonomous car race work?
Autonomous vehicles are tested on a track delineated by cones, with yellow ones on the right and blue ones on the left. The in-car system uses those different colours to figure out whether it needs to steer left or right to stay within the track boundaries.
Ampera’s car uses a camera to detect the environment, then asks Raspberry Pi for help to build a map that allows the humans on the team to decide the optimal path for it to complete the track. The camera is a Luxonis OAK DW with an integrated vision processing unit (VPU), which allows it to run the YOLO object detection model to “see” where the cones are on the track.
The camera is connected directly to a Raspberry Pi 3 Model B, and the team stays remotely connected to the Raspberry Pi via Wi-Fi to allow for real-time debugging. Once the Raspberry Pi receives the information about the cones, the human team members run their Python algorithms to build a map of the track. These algorithms don’t just calculate the best path to follow; they also determine control actions, such as steering and acceleration, and send these suggestions to the car’s electronic control unit (ECU). The ECU controls the motor of the car, using the standard CAN bus communication protocol. The Raspberry Pi is connected to a bespoke PCB with a CAN transceiver and a controller module which allows it to do all of this, as well as talking to the camera it’s also directly connected to.
Thanks to Gabriel Ferreira Gelain for sending over all the photos and videos of their driverless cars, and congratulations to the entire team for everything they’ve achieved so far. You can follow Team Ampera on Instagram, LinkedIn, Facebook, and YouTube to keep up with their electric dreams.