NERVZ built a 1/10-scale self-driving race car for the F1Tenth competition, aiming to complete laps as fast and reliably as possible on unknown tracks. The project focuses on full system design, from sensing and decision-making to control and embedded implementation.

The system is split cleanly. An NVIDIA Jetson Orin Nano processes LiDAR data and runs our custom-designed algorithm, Weighted Pairs MMSE with Sharp-Turn Detection, to generate speed and steering commands. These are sent to an STM32 microcontroller, which handles fast, deterministic control of the motor and steering. This separation keeps high-level computation and low-level control efficient and tightly integrated.