a. Prototype
For our evaluation, we decided that that the functionality we want to test, is the ability of the Dogg0 (Miro-E) robot to follow the PwD during their walks. The walks will take place in the garden of the nursing home and are assumed to be flat.
In the final product, Miro-E needs to have the following capabilities while going for a walk:
- Follow the PwD using it camera sensors
- Navigate through the garden and avoid obstacles with sonar sensors.
- Interact with the PwD by making dog-like noises and movements.
For our evaluation we decided to focus on one functionality, namely using the camera sensors to follow the PwD. For simplification purposes there is assumed that the PwD is the only person in the view of the Miro-E and that there are no obstacles present.
The Miro-E has two front cameras with a resolution of # x # pixels and a combined view of approximately 160 degrees. This enables the Miro-E to follow the PwD right behind him/ her. A lightweight person detector model will be used to locate a boundary box within the camera frame, like YOLO from openCV. We then will create an algorithm and simple controller that uses this information to keep this boundary box in a certain location in the frame by driving the wheels.