Wheel speeds not variable

[jim-v]

It has been observed that when doing a slight turn, the robot goes straight. Note that this is independent from any mechanical issues.

DOD: can go in the eight directions in keyboard teleop.

[jim-v]

The issue is not in the ROS node. m in teleop_twist_keyboard produces
x -0.500000 theta -1.000000 delta -0.100000 left -0.400000 right -0.600000 in cmd_vel_callback

[jim-v]

If I modify https://github.com/UbiquityRobotics/loki_base_node/blob/turn/bus_server.py#L282
to have * 2 instead of / 2, then it does move in the correct way. I don't think this is a fix, but it does give some information: velocities of 0.4 and 0.6 are the same by the firmware.

[waynegramlich]

I reread a bunch of the code. I added a description of the line protocol to bus_server.py. That way you can play with Loki with a serial line program line minicom. It should be possible to shut down the bus_server node and run minicom directly on the Loki.

The firmware code is needlessly complex right now, since Freya is no longer supported. All of the firmware finally lives in a single repository (bus_loki) in the rev_f directory and a few sub-directories. Eventaully, all of the code should be migrated into a single .ino file, but not today. The code has a dependency on Arduino and Arduino-Makefile repos.

[rohbotics]

Why if statements?

Can't you just do left = x - (theta * wheel_spread * gear_reduction) and right = x + (theta * wheel_spread * gear_reduction) for all the cases?

[waynegramlich]

I think the if/elif/else stuff is left over from the original ROS Arduino Bridge code originally written by Patrick Goebel. If I had written the code, I probably would have computed wheel_spread * gear_reduction as a constant that does not change.

[waynegramlich]

@jim-v can you rename the issue more descriptively. I think you said it wasn't a cross-over issue at the huddle.

[jim-v]

Renamed the issue to clarify that it is not a cross-over issue. Straight forward and reverse and rotate in-place work, but the diagonals in teleop-twist-keyboard do not. The commands sent from the ROS node do appear to vary correctly, so it looks like a dividing issue in the firmware.