Smooth Operator

I built Smooth Operator, an autonomous driving system for the F1 2020 game. I trained a deep neural network on synchronized gameplay frames and F1 telemetry to predict future racing trajectories and target speeds, which are then converted into real-time steering and braking controls.

I built the complete pipeline including data collection, telemetry synchronization, computer vision, neural network training, Bézier trajectory prediction, physics-aware loss functions, model evaluation, and real-time game control using Python, PyTorch, OpenCV, and NumPy.