Create basic experimental setup to run and obtain data

[chcorbato]

Just 2 workstations (goals) - try with the factory map (desired), but if problematic use the map we have.

Should follow the definition in https://github.com/rosin-project/metacontrol_nav/issues/4

Needs:

• [x] Data logging tool #25
• [x] Create factory environment
• [x] Define Goal positions - a set of 3 values
• [x] Define Initial positions - a set of 3 values
• [x] Create map - only one is going to be used
• [x] Define launch configuration: - a set of 3 configurations (Fast / Standard / Safe)
• [x] Create control script for the experiments: #29

[marioney]

Factory environment is created in #27 and the factory map is created in https://github.com/rosin-project/metacontrol_nav/commit/8f57deac1e2f94a67ba1e3ac827c163b59cf961b

[marioney]

3 Goal positions and 3 Start positions are defined in the image

@chcorbato @darkobozhinoski does theses position seem ok to you?

[chcorbato]

3 Goal positions and 3 Start positions are defined in the image

@chcorbato @darkobozhinoski does theses position seem ok to you?

Looks good to me!

[chcorbato]

In the current setup: metacontrol_sim:https://github.com/rosin-project/metacontrol_sim/commit/6d9e3d015e202988e3b6a793531e4b1be37d71d6 metacontrol_nav:https://github.com/rosin-project/metacontrol_nav/commit/fa078c4f0c39f87c69fc4ed6de1fe2702f646197

The metacontrol correctly triggers adaptation (see below Observation energy above 1.5 threshold) changing from the departing configuration fd_navigate_fast to fd_navigate_safe, and resumes navigation towards current goal, but gets stuck (see image) fg_nav_fast None [('obs_energy_1', 2.04421472549), ('obs_energy_2', 1.65799868107), ('obs_energy_3', 2.33954739571), ('obs_energy_4', 1.47190856934), ('obs_energy_5', 1.46199381351)]

[marioney]

resumes navigation towards current goal, but gets stuck (see image)

I've just checked and I have the same error. Seems that the local planner is ignoring the global plan. I'll try to fix that

Edit:

The issue was the minimum in place rotation velocity, it seems that wasn't high enough to turn the robot. I changed it from 0.314 to 0.35 in https://github.com/rosin-project/metacontrol_nav/commit/eb52af2cbab50c531537997500ebeda55df7be02 and it works for me.

@chcorbato can you test with that value and report back.

[chcorbato]

@chcorbato can you test with that value and report back.

It works for me too, also with metacontrol adapting half-way. Do you want me to help testing with other start and goal positions?

[chcorbato]

In relation to:

• Define launch configuration: - a set of 3 configurations (Fast / Standard / Safe)

Is this done/ready?

Additionally: I think that currently configurations correspond to different values {'safe','standard','fast'} for the argument profile in the launch file: metacontrol_nav/launch/include/move_base.launch

But in principle @ipa-nhg is generating different launch files for each configuration. We need to fix a way to proceed with this, because it affects the implementation of rosgraph_manipulator

[marioney]

I think that currently configurations correspond to different values {'safe','standard','fast'} for the argument profile in the launch file:

You're right, these are the configurations.

But in principle @ipa-nhg is generating different launch files for each configuration. We need to fix a way to proceed with this, because it affects the implementation of rosgraph_manipulator

Does the reasoner need to know the current (initial) configuration beforehand? Or can we start with one of those three and perform reconfiguration then? In any case, we can use the launch files generated by @ipa-nhg, we just need to include them onthe metracontrol_sim main launch file.

[chcorbato]

Does the reasoner need to know the current (initial) configuration beforehand? Or can we start with one of those three and perform reconfiguration then?

In any case, we can use the launch files generated by @ipa-nhg, we just need to include them on the metracontrol_sim main launch file.

That is ok, but I think this only affects how the experiments are deployed. There are two uses of these files:

• metacontrol uses them as resources to execute adaptations. For this it is important to standardize where(folder) are those files located and how they are named.
  • where do you think we should have them in metacontrol_sim or in metacontrol_nav? @marioney @ipa-nhg
• deploy the navigation application for the experiments (this is what you mention in your comment). Ideally, we do not have the files in two places, but just use the same ones.

[chcorbato]

Setup ready, obtaining the data is https://github.com/rosin-project/mros1_reasoner/issues/10#issuecomment-630073391