R16-U3 · usually age 16 · Autonomy Innovators
Machine Learning Literacy
When is ML useful, and when does it fail?
Outline: sessions not yet written
Build: Toy object classifier
- Sense
- Test and debug
Who it is for
Age is a starting point, never a limit. A learner who can show what this unit starts from can begin here at any age.
- How they learn
- By designing systems in C++ or Python, using simulation as preparation, and proving behaviour with measured trials.
- The adult's role
- An adult reviews safety, tools and higher-risk subsystems.
- Session
- 90–120 minutes
- Hardware
- Arduino, Raspberry Pi Pico, ESP32 or a Raspberry Pi computer. Many of these boards are 3.3 V: use level shifting for 5 V parts. Advance only after driver and power competency.
Starts from
Vision
Use camera images as sensor data and measure how reliable detection is.
Built in R16-U1 Images as Sensor Data
Data-driven tuning
Use logged data and graphs to tune a system.
Built in R14-U3 Data and Tuning
Help or harm
Say who a machine could help or harm, and how to make it fairer and safer.
Built in R09-U4 Smart Crossing
Builds, and what it is the input for
ML literacy
Train and test a simple classifier, and find where it fails.
Input for R16-U4 Sort for a Purpose
Responsible robotics
Weigh safety, privacy and fairness in how a robot is used.
Input for R17-U4 Community Robotics Capstone
In the studio today
No studio lab or activity for this unit yet.
Outline: sessions not yet written
This unit is an outline: a title, a question and an intended prototype. Its sessions, materials and adult guide are not written yet.
- Ideas
- training/test separation · false positives · bias
- Evidence of learning
- Compare failures on an unseen test set.