UM-Model 1 understands the physics of driving—not just patterns. It reasons about causality to make safe decisions in novel situations where pattern-matching AI fails catastrophically.
Why causal understanding prevents accidents
Pattern matches: "In training data, braking worked 70% of the time in similar situations"
Apply brakes ✗
Fails in novel situations. Doesn't understand WHY braking works (physics, road conditions, speed). Catastrophic failure when distribution shifts.
Causal model: Speed → Stopping Distance, Road Friction → Braking Effectiveness, Mass → Momentum
Simulates both actions: P(Collision|do(Brake)) vs P(Collision|do(Swerve))
Swerve (wet road + high speed = insufficient braking distance) ✓
Explore causal decision-making in autonomous driving
Proven results from autonomous vehicle testing
Causal understanding enables safe decisions in situations never seen during training
Causal models predict outcomes faster by understanding physics, not searching patterns
Understands how weather causally affects friction, visibility, and stopping distance
Predicts pedestrian intent through causal reasoning about human behavior and context
Test counterfactual scenarios with causal analysis
Adjust parameters and click "Run Analysis" to see decision predictions