Design a machine that sorts objects into two bins by what a sensor sees: a light sensor (or a trained model) decides, a servo flap acts, and accuracy is measured on a test set.
Recycling centres sort millions of items a day with sensors and flaps. Can yours sort a box of mixed items, and how often is it right?
Client: the school's recycling team. Context: light and dark items (or paper and plastic) slide down a chute past a sensor.
Brief: a light sensor reads each item; light items (reading 400 or more) go to bin B, dark items to bin A; a servo flap moves within 0.5 s; the sorter must be at least 90% accurate on a test set of 20 items. Kit parts only; USB power.
Inquire: Read the brief: who it is for, what it must do (criteria) and the limits (constraints).
Develop: Plan before you build: parts, circuit, code and the tests you will run.
Create: Build it, first in the simulator, then for real if you have the kit. Expect a few tries; that is normal.
Test: Run your tests and record what happened, including what failed.
Evaluate: Judge it against each criterion, and say what you would change.
Make your prediction first. Help opens once you've locked it in.
InquireRead the brief: who it is for, what it must do (criteria) and the limits (constraints).
Inquire: the brief
Sort each line of the brief: a criterion (what it must do, which you can test) or a constraint (a limit you must work within)?
Light items (400 or more) go to bin B
The flap moves within 0.5 s
At least 90% accurate on 20 test items
Kit parts only
USB power
DevelopPlan before you build: parts, circuit, code and the tests you will run.
Develop: plan before you build
Stretch route: replace the light sensor with the Teachable Machine classifier from Code › micro:bit and the computer, sending its label to the board over serial.
Try it yourselfYour edits are saved in this browser.
Here the LED stands in for the flap (on = bin B). Write the decision so all four tests pass.
Use the board. This ticks when you get there.
Your sorter puts 17 of 20 test items in the right bin. What is its accuracy, as a percentage?
CreateBuild it, first in the simulator, then for real if you have the kit. Expect a few tries; that is normal.
Create: build it
TestRun your tests and record what happened, including what failed.
Test: run your table
EvaluateJudge it against each criterion, and say what you would change.
Evaluate: against the brief
For teachers
A Years 9–10 design project: an automated sorter where a light sensor (or a trained model) decides and a servo flap acts, prototyped on the board, built, and measured for accuracy on a 20-item test set. About 250 minutes.
I can turn a sensor reading into a decision and an action.
I can measure accuracy on a test set.
I can explain why a design that passes its code tests may still sort badly.
Before class
Real light readings depend on room light and how the item passes the sensor; a shade or an LED lighting the item helps. The Servo library is a recipe until the class has met libraries. Links: Code › Train a recycling sorter (Teachable Machine) for the stretch route.
If you assess with MYP Design, the steps line up with the criteria:
A Inquiring and analysing: the brief, its criteria and constraints, and research into parts (Inquire).
B Developing ideas: a design specification, the circuit and code plan, and the test table written before building (Develop).
C Creating the solution: building it (in the simulator, then for real), following the plan and justifying changes (Create).
D Evaluating: testing against the specification, recording failures, improvements and impact (Test, Evaluate).
Otherwise, use the same four headings as a simple rubric: each step's reflect checks are the evidence list.
Running the lesson
Develop: the code tests and the accuracy test answer different questions; ask which is which.
Test: someone else chooses the 20 items.
The hard part (on purpose)
Students treat 100% on their own five items as proof. Hold the 20-item test set back until Test.
When students get stuck
Accuracy is poor. Ask: What do the misses have in common? Fix: Calibrate the threshold with real items, or shield the sensor from room light.
Stretch and support
Stretch: The Teachable Machine route: compare its accuracy with the light sensor's on the same 20 items.
Support: A servo recipe card and a results table for the 20 items.
Where it leads
Other Years 9–10 projects: Two-motor drive base, Lift.
What each step is for
Inquire: the brief: Separating what the design must do (criteria) from the limits it works within (constraints) is the first move of any design brief.
Develop: plan before you build: Prototyping the decision against threshold tests, then planning a separate accuracy test with real items, separates "the code is right" from "the machine sorts well".
Create: build it: Building in the simulator first catches wiring and code slips cheaply; the real build then tests the design against the world.
Test: run your table: Running the test table written in Develop, and recording every result including failures, is the evidence for the evaluation.
Evaluate: against the brief: Judging the design criterion by criterion, with evidence from the tests, is what separates an evaluation from a description.
Now you can
Turn a sensor reading into a decision and an action.
Measure accuracy on a test set.
Explain why a design that passes its code tests may still sort badly.