Ros2 Computer Vision Controllable Runtime

[Cruiz102]

In the migration of ROS1 to ROS2 the first module we are going to migrate is the Computer Vision Publisher. In the same time I want to make a more modular computer vision node that can be controllable trhough the hydrus-cli.

The idea im having is to make all the scripts that we have available to the cli.

• Something I want to support in this CV node. Is the generality of accepting multiple cameras inputs.

• Im thinking making a camera abstraction for all the cameras that we would want to support.

• Change the pipeline so that if we have a 3d depth map avlaible and the IMU position it automatically calculates the 3d points.

• Implement Strategy pattern to the detector pipeline so that its easier to change detectors algorithms in runtime.