Race to the Future

For the 'Race to the Future' project, our Digital Innovation class was split into two teams, each responsible for developing an autonomous RC car. The goal was to design a self-built car that could navigate a track completely autonomously. The team that won the race had to show not only speed but also precision in avoiding obstacles and following the track.

Functionalities

Our autonomous RC car was equipped with the following capabilities:

• Fully autonomous navigation over the track using a camera.
• Use of a wide-angle lens to provide a broader field of view for better track perception.
• A custom 3D-printed chassis, specifically designed to support the hardware.

Project Execution

We used the Donkeycar library, an open-source platform for building self-driving cars. This was installed on a Jetson Nano, a powerful edge computer mounted on the car.

After installing the Donkeycar library, we started collecting data by manually driving the car over the track. The collected data was then used to train an AI model with TensorFlow and Keras. This model enabled the car to make autonomous decisions and follow the track.

Our code was entirely written in Python, using OpenCV for image processing. The combination of hardware and software allowed our car to avoid obstacles and efficiently navigate the track.

A crucial decision was adding a wide-angle lens to the camera. This significantly improved the car's view, enabling it to react faster and more precisely to the track and obstacles.

Conclusion

Thanks to our strategy, including the choice of a wide-angle lens, a well-trained AI model, and a sturdy 3D-printed chassis, our team won the race. This project was a valuable experience in combining hardware and AI technology. It emphasized the importance of data analysis, teamwork, and innovation. The results were more than satisfying, and the success of our autonomous car gave a glimpse of the possibilities of AI in mobility.