Tests on the robot

[vvasco]

We want to test the new features introduced and tested in simulation on the robot, specifically the obstacle detection alone and then within the TUG. Finally I will do several tests of the whole demo to make sure everything works properly.

[vvasco]

I updated the default parameters of obstacleDetector and managerTUG in order to:

• allow obstacleDetector to detect legs (few points are detected, thus I lowered the threshold on the minimum number of points inside a cluster to be valid); this might generate more false positives, thus managerTUG now discards obstacles if they don't occur within a certain frequency;
• reduced the radius for stopping navigation if an obstacle is encountered (from 1.5 m to 1 m), since walls behind the robot easily fall inside this radius.

Obstacle detection out of the box

In this experiment, I ran the obstacle detection out of the box. The detected obstacles in the middle viewer are the walls, the table legs and my legs. When commanding the robot to reach 2.5 m along the x direction, it stops before because the legs fall into the radius of 1 m. We can verify this with the get_state command, whose output tells us that the robot has reached 1.5 m.

Obstacle detection within the demo

In this experiment, I put an obstacle (a box) along the robot's path. The robot stops and asks to remove the obstacle. When this is removed, the interaction starts again. Here a (rudimental) video showing the functionality (I need to make a proper one).

[pattacini]

Closing in favor of #306.