search

cmu-mars / issues

This is a metarepo for issues, milestones, and project management.
0 stars 0 forks source link

IG Graph has wrong ordering #107

Closed schmerl closed 6 years ago

schmerl commented 6 years ago

(For some reason this is only happening when aruco-camera is the start configuration.) When changing into a new configuration away from aruco and towards mrpt, the following instruction graph is generated:

Recieved a new goal: P(V(1, do SetReconfiguring(1) then 2),V(2, do KillNodes(%aruco%) then 3)::
V(3, do KillNodes(%mapServerObs%) then 4)::V(4, do SetSensor (%KINECT%, %on%) then 5)::
V(5, do StartNodes(%mrpt%) then 6)::V(6, do StartNodes(%mapServer%) then 7)::
V(7, do StartNodes(%laserscanNodelet%) then 8)::V(8, do SetReconfiguring(0) then 9)::
V(9, do MoveAbsH(40.54, 16.97, 0.35, -1.5708) then 10)::
V(10, do MoveAbsH(40.54, 12.92, 0.35, 3.1416) then 11)::
V(11, do MoveAbsH(38.97, 12.92, 0.35, -1.5708) then 12)::
V(12, do MoveAbsH(38.97, 8.38, 0.35, 3.1416) then 13):
:V(13, do MoveAbsH(36.27, 8.38, 0.35, -1.5708) then 14)::
V(14, do MoveAbsH(36.27, 0.00, 0.35, 3.1416) then 15)::
V(15, do MoveAbsH(22.70, 0.00, 0.35, 3.1416) then 16)::V(16, end)::nil)

My current theory is that the reconfiguration actions are in the wrong order. They should be:

...SetSensor(%KINECT%, %on%), StartNodes(%laserScanNodelet), StartNodes(%MapServer%), StartNodes(%mrpt%) Not sure where this is done -- in the planner or in the translator.

schmerl commented 6 years ago

I no longer think this is the cause of the problem. I think the problem was that the state (i.e., the pose) needs to be published if AMCL or MRPT start. This is now happening in the TA (which is not the best place for it, but hey).

Keeping it open to run more tests.

schmerl commented 6 years ago

This turned out not to be the issue. It was an issue with setting the initial_pose that is required by MRPT and AMCL.