Closed jack-oquin closed 6 years ago
Commit d3eb4c6 removes the hokuyo_node
build dependency.
It still needs testing on a real segbot_v2.
Commit b2e47ae works around the Qt problem with building libsegwayrmp (segwayrmp/libsegwayrmp#17),
That allows the segbot_v2 base driver to compile. It still needs testing on a real robot.
Further testing requires installing Xenial and Kinetic on one of the segbot_v2 robots. Since it's unlikely to work right off the bat, the ability to dual-boot both Xenial and Trusty is required.
@nickswalker installed Xenial on the arm robot, and was able to get it moving with teleop. That actually proves quite a bit about libsegwayrmp and segway_rmp.
He was unable to do full navigation, because that robot currently has no display attached, making it difficult to localize and debug any navigation difficulties.
I am now optimistic that the other segbot_v2 machines will come up and run with only a few problems, once Xenial and Kinetic are installed.
Bender has Xenial and Kinetic installed and is currently executing visit_door_list.
The navigation looks a bit jerky, with a tendency to over-steer.
The first thing to investigate is the laser output, since we now use urg_node on Kinetic in place of the deprecated hokuyo_node.
Bender ran visit_door_list for well over an hour today, drunken steering and all.
All the segbot_v2 robots are installed, configured, the software builds and the robots run basic regression tests.
Note that there are some problems with navigation getting stuck, see: #99.
Let's build the
segway_rmp
package in our source tree.It depends on segwayrmp/libsegwayrmp#17. That package uses cmake, not catkin. In order to build it in our source trees, we'll need to use
catkin build
rather thancatkin_make
. Looking at the issue above, we need to migrate some more Qt4 code to Qt5, similar to the utexas-bwi/bwi_common#95 tasks.All references to
hokuyo_node
must be removed and migrated tourg_node
.Then, the segbot_v2 laptops need to have Xenial installed.