Currently i am looking at implementing detection of obstacles within a certain metres away from a stereo camera feed. I am currently using a third party USB stereo camera and have written a ROS node to images and send them in ROS as sensor_msgs/Image topics.
I also realized that there are 2 different isaac ros packages that works with Stereo cameras: isaac_ros_ess and isaac_ros_bi3d
would like to find out:
I see that default example produces a point cloud, in the case of trying to calculate obstacle distance to the camera in real world units, do i need to use DisparityToDepthNode from isaac_ros_stereo_image_proc? i have tried doing so but no sure what units is the depth image supposed to be in.
For disparity_values for isaac_ros_bi3d, how do i find out what are the valid range of values? and how is it related to the real world? also, is max_disparity_values representing the number of discrete levels to segement into
welcome to the Isaac ROS forum and thank you for your questions.
Let help you line, question by question:
Have you also evaluate to use Isaac ROS pose estimation?
The packages collected use GPU acceleration for DNN inference to estimate the pose of an object.