Post #9

April 29, 2025

Here, we started testing the perception capabilities of the Stretch robot. First, we visualized the robot's point of view with depth information through point clouds, and we were also able to segment and label various objects in the room with deep learning models. In the context of our project, we can see in the first image that Stretch is able to perceive its surroundings with depth as a point cloud, and we can see in the second image that the Stretch is able to identify a chair which our patient would be sitting in. 

Next, we tried out AruCo marker detection for more precise object detection. Here, we placed an AruCo marker on a chair and verified that the robot was able to detect the marker as "target_object". In the context of our project, this would enable the robot to navigate towards the AruCo marker as a designated target and orient itself properly before beginning an exercise routine with a user.