JetRacer — High-Speed AI Racing Robot

JetRacer — High-Speed AI Racing Robot

JetRacer is an open-source AI race car built on the NVIDIA Jetson Nano. It uses a 1/10th-scale RC car chassis with a Jetson Nano mounted on top, enabling neural-network-based road following at speeds exceeding typical educational robots. Steering is controlled by a servo and throttle by an ESC, both commanded directly from the Jetson via I²C.

JetRacer ships with Jupyter notebook examples for data collection, training, and live inference of a road-following network (a ResNet18 regression model predicting x/y deviation from center).

Hardware

• NVIDIA Jetson Nano Developer Kit
• 1/10-scale RC car chassis (e.g. Waveshare JetRacer Pro)
• Front-facing wide-angle camera
• PCA9685 PWM servo controller
• 7.4V Li-ion battery
• 5V power bank for Jetson

What It Can Do

• Road following — ResNet18 regresses steering angle from a single camera frame at inference-time
• Object following — detects and tracks objects using SSD MobileNet
• Live teleoperation — gamepad-driven control via Jupyter widgets
• Data collection — annotate frames in-browser by dragging a target point

Source

• Repo: https://github.com/NVIDIA-AI-IOT/jetracer
• License: MIT