ED sensor integration

Plugins

ED Kinect plugin

Compares recent sensor image with an image rendered from the world model. Based on this comparison:

• the pose of existing entities is updated
• new entities are added
• entities are removed from the world model

Interfaces:

• foo
• bar

ED Laser plugin

• ed_laser_plugin

Interfaces:

• foo
• bar

ED Clearer plugin

• ed_clearer_plugin

Interfaces:

• foo
• bar

Tools

Several tools are available for testing the features of this repository. These can be found in the [tools](tools) along with their documentation.

Issues with current toolchain

• The json resulting from ed_image_saver only contains sensor pose, rgbd filename and timestamp, not the world model. This makes sense in a way (it saves images, not the world model). However, the segmenter tool cannot be used easily because the right data is not present.
• Updater.update is an long messy function, making it difficult to test standalone parts.
• No unit tests
• World model cannot be serialized entirely. Therefore it's difficult to create realistic situations.

Approach

• Create a simple world model (check, e.g., this test for inspiration) in a unit test containing a piece of furniture
  • Move the piece of furniture
  • Render an image from this (for inspiration, check the fast simulator for inspiration)
  • Try to update the entity pose based on this image. When does/doesn't it move?