AI & TinyML / TEACHING GUIDE
Teach a robot to recognize the road
From camera images to collision avoidance.
A teaching sequence
- Start with control
Bring up the robot, verify camera and motor operation, then try basic motion and teleoperation before adding a model.
- Build a dataset
The class agenda called for 100 good FREE images and 100 BLOCKED images. Discuss consistent labels and variation in the scene.
- Train and transfer
Use the compatible training notebook in Colab, export the model, and run the collision-avoidance demonstration on the robot. Match model filenames and library versions to the notebook you actually use.
- Test and extend
Test in a clear, supervised space. Record incorrect predictions and changes in lighting or background. The course continued into line/road-following notebooks.
Source & adaptation
Reconstructed teaching outline from the January–February 2026 daily agendas; use the linked platform documentation for installation and code.
Use this as a planning guide and adapt it to your equipment, students, and timetable.
Find handouts and platform resources →
Original AIM teaching material may be adapted with attribution to David McLoda. Reuse details