Skip to content

Lab 4 · AI Inside & Out — Hand-Gesture Motor Control

So far your motor programs have been hard-coded: they always do the same thing no matter what is happening around them. This lab changes that. Instead of fixed instructions, the motor responds to your hand movements — detected by a camera and a machine learning model — without you changing a single line of code.

Step 1 · Plug in the webcam

Connect the USB webcam to your Raspberry Pi before starting.

Step 2 · Open the file

cd hotis-projects-starter
geany ai-motor.c

Step 3 · Set up the motor pins

In the setup() function, you need the same three pinMode() calls you used in Lab 3. → pinMode() · Pins & constants

Step 4 · Understand the AI data

In loop() you will see a function call:

get_handstate(fp, &lHand, &rHand);

This fills lHand and rHand with the latest data from the hand-tracking AI. Each HandState has four fields: x, y, theta, and mag.

→ See Hand state (AI) for the full documentation on get_handstate() and how to access fields.

Step 5 · Explore what the fields mean

Before you try to control the motor, run the program and watch what prints:

make ai-motor
./ai-motor

A webcam window will open. Hold up your left hand and move it around. Watch the terminal.

Experiment: The printf at the bottom of loop() prints one field — try swapping theta for mag, or x, or y. Recompile and run each time to see what changes.

Your exploration table

Fill this in from your own observations (the table is in your paper handout too):

Field What it seems to measure Range you observed
lHand.theta
lHand.mag
lHand.x
lHand.y

This table is yours to fill in

The answer is not written here — that is the point of this step. What you write in that table is your discovery from running the program and observing it.

Step 6 · Control the motor with your hand

Now you know what the fields mean. Use that knowledge to write the loop() function so your hand controls the motor.

Some things to keep in mind:

Speed scaling: The motor accepts values from 0 to PWM_MAXVAL (1024). Hand angles (theta) are roughly in the range −180 to 180. Those ranges do not match directly — you will need to scale. → Scaling numbers · pwmWrite()

Direction control: The two direction pins must always be opposite (HIGH/LOW). You cannot set them proportionally — they are either one state or the other. You will need an if/else to decide which state based on some hand value. → Making decisions (if/else) · digitalWrite()

You are the designer: Which gesture controls speed? Which controls direction? The handout does not decide for you. Pick something that makes sense to you, try it, and adjust.

What to try next

  1. Can you control the motor speed with your hand? (Start here.)
  2. Once that works, can you control both speed and direction using different gestures?

See also