Students will
- Run the pre-trained Glass Breaking Sensor example without changing its model
- Collect repeated predictions from contrasting audio samples
- Explain one limitation, possible bias, or unsafe use of the classifier
Arduino UNO Q 2GB
Test a pre-trained audio classifier
Run a reviewed App Lab audio-classification example, test it with teacher-provided sounds, and use evidence to describe where the model succeeds and fails.
MS-ALG-PS-04Algorithms & DesignMS-DAT-DI-27Data & Analysis
Evidence of success
- A completed results table with at least six trials
- A repeated test of one ambiguous sample
- A claim about model performance supported by observations
- A written limitation and a justified boundary for use
Hardware
- Arduino UNO Q 2GB
- USB-C data cable
- Computer with Arduino App Lab
- Teacher-provided non-sensitive audio samples
Prerequisites
- Connect UNO Q and run an App Lab example
- Describe a prediction as a model output rather than a fact
- Follow classroom privacy rules for recorded media
Before class
- Run the current Glass Breaking Sensor example before class
- Prepare a small set of licensed or teacher-recorded audio files that do not identify students
- Decide whether students may record any audio; default to teacher-provided files
- Prepare a simple results table with sample, predicted class, confidence, and observation
Build and test
- 01Open the Glass Breaking Sensor example from App Lab Examples and read its description before running it
- 02Run the unchanged example and confirm that its browser interface loads
- 03Test at least six teacher-provided clips across more than one sound type
- 04Record the predicted class, displayed confidence, and whether the result matched the team's expectation
- 05Repeat one ambiguous or surprising sample three times
- 06Summarize where the model worked, where it failed, and why its output should not be treated as certainty
- 07Propose one low-stakes classroom use and one use that should be avoided
Optional engineering record
Record evidence from this lesson
Choose the prompts that help students explain predictions, evidence, debugging, and transfer. Saving creates a new entry in this browser’s Rudi notebook.
Record your thinking — optional4 prompts
Use these prompts if they help students capture evidence, decisions, or questions. You do not need to complete every prompt or create an entry at every step.
Nothing is saved until you choose this button.
Troubleshooting
The example interface does not load
Confirm the UNO Q setup lesson works, restart the unchanged example, and check the current App Lab example instructions before modifying files.
Every sample receives the same prediction
Verify the files play correctly, use clearly contrasting samples, and record the failure as evidence rather than assuming the model is correct.
2026 middle school standards
Potential CSTA connections
These are evidence-based crosswalk candidates for curriculum review—not a claim of official alignment.
Justify whether a problem is best solved using procedural instructions, rule-based logic, data-driven methods, or a combination of these approaches.
Students compare a data-driven classifier with rule-based or procedural approaches and justify an appropriate boundary for use.
Lesson evidence: Justified low-stakes use and avoided useSummarize a data investigation process, including potential biases, limitations, and supporting evidence.
Students summarize repeated classification trials, including limitations and supporting evidence.
Lesson evidence: Six-trial results table and repeated ambiguous testSupporting concepts
sketch-foundations
Sources and review
Arduino UNO Q documentationArduino App Lab documentationArduino App Lab examplesStarting point, not verified curriculum. Review the actual hardware, circuit, code, power requirements, and classroom conditions.
After teaching this lesson