7 · Sensors + AI together (and a practice drill)
One sensor with AI is useful. Several sensors with AI is where robots get genuinely capable — because each sensor covers the others' blind spots.
Think about why a self-driving feature uses more than a camera. A camera is great in daylight but struggles in glare or fog. A distance sensor doesn't care about light but can't read a stop sign. Combine them — let the AI weigh a camera image and a distance reading and a microphone — and the robot can handle situations none of them could alone. Engineers call this sensor fusion: blending several senses into one better decision, the way you use eyes and ears to cross a street.
You can build a tiny version of this with the boards you already know. The micro:bit alone carries an accelerometer (motion), a microphone (sound), a light sensor, a temperature sensor, a compass, and buttons (Micro:bit Educational Foundation, n.d.-b) — train a model on the motion sensor with CreateAI, then combine its guess with, say, the light sensor in your code. The Arduino path adds machine learning right on the microcontroller via TensorFlow Lite, so your own attached sensors can feed a learned model on the device (Arduino, n.d.).
The big idea to carry forward: more, varied senses → a more reliable "think" step. And the flip side, which sets up the rest of the course: even with many sensors and a trained model, the robot is still making a guess — so we must decide, on purpose, what it's allowed to do on its own and when it must stop and ask a human.
Drill below. Lock in Parts 1–4 before we get to safety and ethics.
Sources
Micro:bit Educational Foundation. (n.d.-b). micro:bit features overview (sensors). https://microbit.org/get-started/features/overview/ Arduino. (n.d.). About Arduino (ML on microcontrollers, TensorFlow Lite). https://www.arduino.cc/en/about
Fill in the missing word from Parts 1–4. Spelling is forgiving — the explanation shows the idea you should have.
1. Adding AI to a robot mostly upgrades one step of the sense → think → act loop: the ___ step.
2. When a person writes out if/else instructions in advance, that's programmed ___; when the machine finds the pattern from examples, that's learned behavior.
3. Teaching a computer to make sense of pictures and video — like recognizing a stop sign — is called computer ___.
4. A learned model's guesses are only as good as its ___ — leave situations out and it may not recognize them.
5. In micro:bit CreateAI you collect movement data, then train, test, and ___ the model to make it better.
6. Blending several sensors so the AI makes one better decision — covering each sensor's blind spots — is called sensor ___.