[WIP] Nav observer

[hsd-dev]

An observer for turtlesim's sensor data. The observer initially trains a model with pre-labeled data and uses the current sensor data to predict the actions turtlesim needs to take to avoid colliding with the wall. The predicted actions are published as diagnostic msg.

nav_model.py code by @ipa-rar

TODO:

• ML code needs python3, melodic running on python2
• sensor data scaling

[hsd-dev]

@ipa-nhg fyi

[hsd-dev]

NavModel exposes a service of type PredictAction and uses roslibpy to avoid issues with Python3.

NavObserver generates a diagnostics msg based on the current sensor values.

level: 0
name: ''
message: "Action prediction"
hardware_id: ''
values: 
  - 
    key: "action"
    value: "Slight-Left-Turn"