In the related Arxiv.org (PDF) the team showcases three different applications of this system for quadrotor navigation:
- flying through a forest
- flying through an unknown shaped/location static/dynamic gap
- accurate 3D pose detection of an independently moving object
They show that their variable baseline system is accurate and robust in all three scenarios.
For the video capture, the Raspberry Pi-based StereoPi board was used. Additional AI-acceleration hardware (Intel Movidus) is considered as a next step, as well as using a more powerful CM4-based version of the StereoPi (v2).
Here is the brief video of the project: