Overview

This project involves creating neural networks and control architectures for autonomous robotic platforms such as autonomous vehicles. We explore various model architectures and machine learning techniques before transitioning to work with a real-world dataset (nuScenes, for example) in order to perform research that is novel and publication-worthy. The eventual goal is to compete in a challenge such as the annual Waymo Challenge.

Current objective

Compete in the 2026 Waymo Open Dataset challenge (end-to-end driving track). Last semester we built models that we submitted to the 2025 leaderboard to gain familiarity with the challenge and the submission process. The challenge timeline has historically run from March to May. We plan to keep refining our models on the 2025 leaderboard, then update to the 2026 version of the competition when it is announced and the dataset is released.

Skills

Fundamentals of machine learning — backpropagation, universal approximation, MLP, CNN, Transformer. Experience with PyTorch or TensorFlow. Understanding of Linux systems and bash.