Option to render learned geometry in video rollouts

[ebianchi]

Contents

The main aim of this PR is to add in the ability to choose whether rollout videos render the predicted trajectories with the original geometry or with the current learned geometry. This decision can be toggled via the flag update_geometry_in_videos in the configuration class SupervisedLearningExperimentConfig.

Tangential to this main aim, the following other minor changes are also incorporated:

• Some hard-coded file/directory locations (a remnant of PR #3 ) are now implemented in a slightly less hacky way. This change is included in this PR because this added functionality requires exporting an updated URDF for every toss video generated. The exported URDF file path from this PR is created in a similar way as the exported video gif file from the prior PR, so it made sense to handle both in a consistent way.
• The color of the true system in the rollout videos is now slightly more transparent. This helped see both the ground truth and the learned system trajectories especially when the learned system geometry is much smaller than the ground truth geometry.
• The example script examples/contactnets_simple.py features setting update_geometry_in_videos = True so reviewers can demo this PR feature.

Call for comments

In this implementation, the MultibodyLearnableSystem extension of the System class now has an additional (optional) argument for its summary method. This argument is the flag for whether to use original or learned geometry in the generated rollout videos. As originally created, this PR does not include any edits to the original System class or any other extensions of it to maintain a consistent set of arguments for summary(). This can be added in if desired, and I welcome feedback on this point in particular.

Examples

The following two gifs are of the same ground truth trajectory (in red) with different predicted trajectories (in blue) as the experiment updates the believed object properties. The geometric updates can be observed in these toss videos.

[ebianchi]

All comments addressed without disagreement.

[ebianchi]

I additionally opened an issue #7 as a reminder to address the current hacked way of getting experiment-level storage locations, so that the issue is remembered beyond the closing of this PR.