Why robots fail

Every robot failure is a failure in one of three places. That's the whole list.

Sense failures

• Dust on a Roomba's cliff sensor → it thinks the floor disappeared → it stops.
• Sun in a self-driving car's camera → it can't read the lane.
• Encoder wire comes loose → robot has no idea how far it has moved.

The fix is usually one of: clean the sensor, add a backup sensor, fuse multiple sensors.

Think failures

• A buggy if-else statement.
• An ML model that wasn't trained on the situation it's now in.
• Latency — by the time the decision arrives, it's already too late.

The fix is usually one of: better algorithms, more data, more compute, simpler logic.

Act failures

• Motor burns out.
• Wheel slips on a wet floor.
• Servo goes out of range and breaks the linkage.

The fix is usually one of: better hardware, gentler control, mechanical safety limits.

The pattern

At every R2BOT debugging session, you ask three questions in order:

1. Did the sensor data look right?
2. Did the brain make the right decision?
3. Did the act actually happen?

This is what professional roboticists do — and now so do you.