Not turning on the spot. about strands_navigation

[gestom]

We tried to play with the DWA planner params and setting the "min_trans_vel to 0.001" and "forward_point_distance to 0.1" (via rosservice call /move_base/DWAPlannerROS/set_parameters) seemed to help in cases when the robot had to turn prior to moving. Try it as well and let me know.

[bfalacerda]

https://github.com/strands-project/strands_movebase/blob/hydro-devel/strands_movebase/strands_movebase_params/dwa_planner_ros.yaml#L30

These are not the most recent parameters. We changed it back to the defaults around the g4s deployment. Probbabliy they got lost with this moving files around...

He turns much more in place when you put these (line 30 and 31) to default.

I dont have time to look into this now, but you can check the commit history in scitos_2d_navigation and look for the correct file

[lucasb-eyer]

On a first impression, changing min_trans_vel to 0.001 seems to be enough. forward_point_distance to 0.1 would be inside the robot! (we measured), why'd you set it that way?

Will also have a look and try out @bfalacerda's mentioned parameters.

[bfalacerda]

@lucasb-eyer it looks like we didnt merge our g4s params upstream. Here is the dwa planner config file that we were using:

https://github.com/BFALacerda/scitos_2d_navigation/blob/with_back/scitos_move_base_params_3d/dwa_planner_ros.yaml

Maybe increasing the sim_time param a bit improves things a bit also. Then, we also discussed the costmap params a bit last year, mainly the footprint - see strands-project/autonomous_patrolling#70
. @AlexanderHermans used this footprint which seemed to improve things:

footprint: [[-0.3000, -0.3200],[ -0.5200, -0.1100],[ -0.5200,
0.1100],[ -0.3000, 0.3200],[ 0.1000, 0.3200],[ 0.2500,
0.1100],[ 0.2500, -0.1100],[ 0.1000, -0.3200]]

The version we're using still has the rectangle footprint being used.

[bfalacerda]

@nilsbore now there are move_base configs in two repos:

https://github.com/strands-project/strands_movebase/tree/hydro-devel/strands_movebase/strands_movebase_params

https://github.com/strands-project/scitos_2d_navigation/tree/hydro-devel/scitos_move_base_params

I'm assuming we're using the ones in strands_movebase. Whats the scitos_2d_nav repo for then, for people that want to use our configs on a scitos without chest cam?

[lucasb-eyer]

I took the time to create and test a much more precise footprint (edit: it also takes care of the screen), which improves navigation quite a bit.

It'd be nice if some of you could also try it out, and if it works, we could use it as a default? Or at least make it a parameter to the launchfile:

footprint: [[-0.30,-0.32], # Back side right
            [-0.52,-0.11], # Back right
            [-0.52, 0.11], # Back left
            [-0.30, 0.32], # Back side left
            [-0.05, 0.31], # Front side left
            [ 0.05, 0.25], # Thick front left
            [ 0.15, 0.13], # Thin front left
            [ 0.20, 0.00], # Front
            [ 0.15,-0.13], # Thin front right
            [ 0.05,-0.25], # Thick front right
            [-0.05,-0.31]] # Front side right

I'm still playing with the other parameters you guys have pointed me to.

[bfalacerda]

this looks good. i'll try it. You don't have problems with the local planner missing the rate (200ms i think) because of the footprint being more accurate?

[lucasb-eyer]

Didn't encounter it yet, but as you know from this morning, I didn't get to see move_base output yet...

[bfalacerda]

comment was in wrong place

[lucasb-eyer]

You sure your last message is in the right issue, and wasn't supposed to be in #137?

[bfalacerda]

closing, more discussion in strands-project/strands_movebase#27