Head mover does not take SEA degree change into account

[SammyRamone]

Expected behavior

Head search pattern should not collide with robot torso or arms.

Current behavior

Search pattern collides a bit.

Steps to Reproduce

roslaunch bitbots_head_behavior test.launch
on branch feature/collision_checking

Context (Environment)

• RViz
• Simulator
• Robot
• Local

Only happens on robot, since the problem is coming from the deviation of the head SEA.

Possible Solution

Adapt search pattern so that lowest line is a bit higher. Then there would be no collision even if the SEA adds a few degrees.

Propably this will be resolved anyway, when the pattern is adjusted to the new camera

[Flova]

I think there is a larger rabithole at this point. Because the transformer and also the normal look_at calls in the head behavior are off due to the sea. Therefore I would introduce a transform in the head part of the tf three right after the head tilt joint. The angels of the new sea joint would be calculated (and published) by a new node which calculates the offset using both Imus.

[Flova]

Of we try to redesign or calibrate the seas that they are only shock absorbers for very high loads and stay quite stiff and centerd while normal use.

[Flova]

Maybe the most precise why would be to add an magnetic encoder inside the sea which estimates it's torsion.

[NFiedler]

@Flova I agree. That would be the best option in the future. But is there a dirty solution which works for now?

[Flova]

I would go for an additional transform for the sea consisting of the difference of the IMU measurements. (To be evaluated.)