Skip to content

Leg detection and following robot controller using the OAK-D stereoscopic camera and AI accelerator.

License

Notifications You must be signed in to change notification settings

hopkira/3dvision

Repository files navigation

3dvision

3dvision.py

Simple program that uses the OAK-D stereoscopic camera and the mobile-net SSD model to detect a person's legs and follow them around the room. Simultaneously, it creates a point cloud that it simplifies into a one dimensional array that tells the robot how close obstacles are that are in front of it. This is work on progress - do not use near small children or animals or in any place where a robot might do damage!

Currently runs at about 10 FPS on a Raspberry Pi 4.

Code will be fully described once complete, but partial descriptions are here:

The program uses logo.py to make the robot move.

logo.py

A simple program that emulates an old fashioned logo turtle and drives the robot using a differential drive (in this case via a RoboClaw motor controller with homemade encoders)

requirement.txt

A list of the Python libraries you will need to get the 3d vision program to run. I'd advise creating a Python virtual environment and "running python3 -m pip install -r requirements.txt"

About

Leg detection and following robot controller using the OAK-D stereoscopic camera and AI accelerator.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages