jeffdacpano28/self-driving-car-rl
🏎️ Train autonomous racing cars with DQN and PPO in a realistic simulator featuring custom physics, interactive track building, and side-by-side performance comparison.
What's novel
🏎️ Train autonomous racing cars with DQN and PPO in a realistic simulator featuring custom physics, interactive track building, and side-by-side performance comparison.
Code Analysis
11 files read · 3 roundsA Python-based simulation environment using pygame that trains and compares DQN and PPO reinforcement learning agents to navigate a track while avoiding collisions and collecting checkpoints.
Strengths
Strong separation of concerns between environment, sensors, agents, and visualization. The ray-casting sensor implementation is efficient and mathematically sound. The dual-screen renderer provides excellent utility for comparative analysis during training.
Weaknesses
Lack of visible unit tests or edge case handling in the core logic (e.g., sensor failures, extreme velocities). Error handling appears minimal, relying mostly on simulation constraints rather than explicit exception management.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
45
Contributors
2
Files
61
Active weeks
5
Repository
Language
Python
Stars
1
Forks
0
License
—