🚗 Real World AI · Sensors + Decision Making

Self-Driving Car Simulator

Explore sensors, write your own if-then driving rules, then switch to an AI learned policy — and discover why you can never write enough rules for every situation!

📡 Sensor Explorer

📋 Write Rules

🤖 AI Policy

⚠️ Edge Cases

🏆 Badge

How Self-Driving Cars Work

📡

Sense

LiDAR, cameras, and radar scan the environment 360° hundreds of times per second, building a 3D map.

🧠

Perceive

AI classifies what each sensor reading means: pedestrian, car, road edge, traffic sign.

🗺️

Plan

Path planning algorithms choose the safest route — balancing speed, safety, comfort, and traffic rules.

🎮

Act

Steer, accelerate, brake — smooth and safe commands to the vehicle's physical control systems.

🚗

Wizzy the AI Tutor

Welcome to the Self-Driving Lab! 🚗 Before a car can decide what to do, it needs to sense the world. The coloured beams are LiDAR rays — they bounce off objects and measure distance. Watch how the sensor readings change as obstacles move closer!

Step 1 — Sensor Explorer

📡 Sensor Readings

Sensors used in real cars:
🔴 LiDAR — laser distance measurement (360°)
📷 Camera — colour, signs, pedestrians
📶 Radar — speed of other vehicles
🛰️ GPS — global position

🚗

Wizzy the AI Tutor

Now you are the AI engineer! Write if-then rules that tell the car what to do. Then test your rules on 5 different scenarios. Can you write rules that handle all of them? Most people discover it's harder than it looks — there's always an edge case!

Step 2 — Write Your Driving Rules

0/5 scenarios passed

📋 Your Rule Set (drag to reorder)

Set up your rules then click "Test All Scenarios"!

Test Scenarios:

Write rules to handle all 5 scenarios. Notice how you can't anticipate every situation with simple rules!

🚗

Wizzy the AI Tutor

Now watch the learned AI policy handle the same scenarios! The AI was trained on thousands of simulated drives and learned to handle situations through experience — not rules. Compare the scores! The AI handles edge cases your rules couldn't.

Step 3 — AI Learned Policy vs Your Rules

Run the AI demo to see how it compares to your rules!

🚗

Wizzy the AI Tutor

The hardest part of self-driving! Edge cases are rare, unusual situations that neither rules nor AI handle perfectly. Click each edge case to test both approaches. This is why self-driving cars are so hard — the real world has infinite situations!

Step 4 — Edge Case Testing

Click an edge case to test both the rule-based system and the AI policy!

🚗

Wizzy the AI Tutor

🎊 You've explored the full self-driving car stack — from sensor beams to learned policies to edge cases! You discovered the bitter lesson of AI: general learned systems beat carefully hand-crafted rules. This insight drives every company from Tesla to Waymo!

🚗

Self-Driving AI Badge!

You explored sensors, wrote driving rules, and compared them to AI policies!

Optional. Stays on this device only — not sent to WhizzStep.

🚗 WhizzStep AI Lab

Activity completion card for

Student Name

has explored Self-Driving AI — Sensors to Policies

This records completion of a browser activity only. It is not an accredited certificate or proof of mastery.

AV Engineer

Sensor Expert

Edge Case Hunter

whizzstep.in

Key Concepts Mastered

LiDAR

📡 Laser Mapping

Fires 128+ laser beams per rotation, measuring distances to build a precise 3D point cloud of the environment.

The Bitter Lesson

🧠 Rules vs Learning

Rich Sutton's famous insight: general methods that learn from data beat hand-crafted approaches — every time, at scale.

Edge Cases

⚠️ The Long Tail

Rare situations that rarely appear in training but must be handled safely. The "long tail" is why self-driving is so hard.

Sensor Fusion

🔗 Combining Inputs

Combining LiDAR + camera + radar gives a richer picture than any single sensor. Each compensates for the others' weaknesses.

Imitation Learning

👨‍🏫 Learn from Humans

Train the AI by imitating how human drivers behave — a key technique used by Tesla's Autopilot system.

Safety Validation

🛡️ How Safe is Safe?

Waymo has driven 20M+ miles. Certifying a system safe enough to deploy requires billions of simulated miles first.

About this lab

AI-25 How AI and Machine Learning Work Classes 8-10 15 min Applied

Learning objective: Steer a simplified simulated vehicle using sensor-style inputs and see how a simple control model reacts.

What this simplifies: A toy 2D simulation is used; it does not reflect the complexity or safety systems of real autonomous vehicles.

Privacy: No learner input leaves the device.

Teacher prompt: Ask the class why this simulation might mislead someone who takes it too literally.

Reflect: What is one thing this activity showed you that you did not expect?

← Hungry Bird Back to All Labs AI Grand Finale →

Bring this learning to my school →