Debug Tool: Simulate Camera Vision

[acbuynak]

Currently unable to simulate the camera vision effectively. A partial implementation exists using a static image publisher in the maze_vision/launch package, but does not work due to a hard coded requirement for synchronous color and depth data for the vision processor.

Ideas

• ROS Bag recording/playback
• semi-static publisher
• synchronous static publishers

Relevant Code

Source of requirement for synchronous color & depth data

maze-runner/maze_vision/nodes/node_maze_vision.py

Lines 60 to 64 in 2c610c9

	sub_color = message_filters.Subscriber(self, Image, "/rs1/color/image_raw")
	sub_depth = message_filters.Subscriber(self, Image, "/rs1/depth/image_rect_raw")
	self.ts = message_filters.TimeSynchronizer([sub_color, sub_depth], queue_size=10)
	self.ts.registerCallback(self._synchronous_callback)

Broken static vision publisher

maze-runner/maze_vision/launch/t1_vision.launch.xml

Lines 11 to 20 in 2c610c9

	<group if="$(var sim_vision)" >
	<!-- Static Image Publisher to stand in for real camera --><!-- HAS KNOWN ISSUES. -->
	<node pkg="image_publisher" exec="image_publisher_node" name="color" namespace="rs1"
	args="$(find-pkg-share maze_vision)/sim/aruco/d1_color.jpg" >
	<param name="flip_horizontal" value="false" />
	<param name="flip_vertical" value="false" />
	<param name="frame_id" value="camera_link" />
	<param name="publish_rate" value="15.0" />
	</node>
	</group>