Wait for your instructor to release the challenge.
Enter the unlock code when it is provided.
Invalid challenge code. Check with your instructor.
READY
Challenge Unlocked
Run the final evaluation to generate your result code.
Your trained agent will run the official evaluation track in
greedy mode — no learning, no randomness.
This gives an accurate picture of what your agent has learned.
ModeGreedy (no exploration)
Q-table updatesDisabled
Time limit90 seconds
Score formulaReward + Completion + Time
Evaluation running… please wait
Your agent completed less than 30% of the track. Train more and try again — or submit what you have.
Challenge Complete
Score
—
Completion
—
%
Time
—
seconds
Collisions
—
Your Result Code
————————————
Write this down or copy it — you will need it for the online submission.
How to submit:
Visit the competition website and enter your name and result code.
You can only submit once, so double-check your name before submitting.
COMPLETE
You trained your first autonomous agent.
You wrote code, observed what the car sees, understood rewards, trained a genuine Q-learning agent, and watched its policy change.
STATE→ACTION→REWARD→LEARNING→IMPROVEMENT
Submit your result code on the competition website to enter the final ranking.