Self Driving Cars
Time required to complete: Around 12-15 50-minute classes, generally 3-4 weeks of instruction
In the Self Driving Car unit, Students engage with a phenomenon about self-driving cars, including a video about self-driving cars on the road and a classroom display of a self-driving car that follows a line, stops for obstacles and encounters difficult decisions that can not be solved without AI. The car encounters a piece of trash and has to decide if it is worth stopping for. Then the car encounters a dog on the track and again, has to decide to keep driving or switch to human-operated mode to navigate around the obstacle.
After this launch, students engage with the decisions that a SDC needs to make to keep us safe and get us where we need to go. This includes covering several AI4K12 learning standards around data collection, training classifiers, neural networks, bias and training data, and human-robot interaction. Students learn about SDCs and then program their own cars to navigate the track and avoid crashing into obstacles. Then students learn about AI features to improve the functionality of their cars, including how to make them fair, unbiased, accurate and reliable in their classification of obstacles.
After training an AI to recognize the difference between a dog and a piece of trash (or slice of pizza), students apply this to their cars and build a car that can operate autonomously but switch to remote human control when needed. Students also engage in activities about how this translates to out-of-school contexts and how AI can support collaboration and things they care about in their home communities. Lessons on comparing a SDC to human drivers and legal/ethical implications (including implications of a crash) are in development.