Back

👆 Click me!

Simulation: Reinforcement Learning

Learn reinforcement learning by actually doing it! Move agents, design mazes, and watch AI learn.

🎮 Interactive Grid World

Control the agent yourself or let the AI learn! Click cells to add/remove obstacles.

⌨��� Use Arrow Keys or Click

Agent Goal Obstacle Path

📊 Live Stats

+0
Rewards
-0
Penalties
0
Net Score
0
Steps
0
Episodes
--
Best Score
+0
Last Reward

🏆 Challenges

🎯
First Goal
Reach the goal once
Speed Run
Goal in ≤10 steps
🔥
Streak Master
3 goals in a row
🧠
AI Trainer
Train AI for 50 episodes

⚡ Types of Reinforcement

Click the cards to see how positive and negative rewards work!

Positive Reinforcement

+10
🤖
🎯

Click to see: Correct action → Reward → Behavior repeats

Negative Reinforcement

-5
🤖
🧱

Click to see: Wrong action → Penalty → Behavior avoided

🎚️ Reward Balance Experiment

See how changing the reward/penalty ratio affects learning behavior

+10
Goal Reward
vs
-5
Obstacle Penalty
Agent Behavior:
Balanced

🎲 Exploration vs Exploitation

Drag the slider to see how the agent's behavior changes!

💡

Exploration

Try new things

Random ε = 0.30 Optimal
🪙

Exploitation

Use best known

💡
🪙
Balanced exploration and exploitation

📈 Watch Learning Happen

See how rewards improve as the agent learns over episodes

Episodes Reward 0 25 50 75 100
Exploration Phase
Learning Phase
Optimal Phase

🌍 Real-World Applications

Click each card to learn more and see an interactive demo!

🎮

Game AI

Chess, Go, Video Games

🚗

Self-Driving Cars

Autonomous Navigation

🤖

Robotics

Motor Control

🛒

Recommendations

Personalized Content

Energy

Smart Grid

📈

Trading

Algorithmic