How do I output trace results in ros format（ros topic）

[shiba24]

Hi @SharineLee , your request to output ros format makes sense. Can you tell me how to you want to save the results?

[SharineLee]

Hi @SharineLee , your request to output ros format makes sense. Can you tell me how to you want to save the results?

I wish this code can be wrapped to a ros node.

If this node is called 'feature_tracker'.

In the testing phase，

I wish feature_tracker can subscribe this topic:

• /dvs/camera_info : sensor_msgs/CameraInfo
• /dvs/events : dvs_msgs/EventArray
• /dvs/image_raw : sensor_msgs/Image

And I wish feature_tracker can publish this topic:

• /feature_tracker/feature : sensor_msgs/PointCloudConstPtr （We call the messages in this topic feature_msg ）
  • // feature_id：indicates which pixels in image frames are from the same feature point
  • int feature_id = feature_msg->channels[0].values[i];
  • // camera_id:left_camera(or mono) = 0 , right_camera = 1
  • int camera_id = feature_msg->channels[1].values[i];// left_camera(or mono): 0 , right_camera: 1
  • // （x,y,z）: Normalized coordinates (using camera parameters)
  • double x = feature_msg->points[i].x;
  • double y = feature_msg->points[i].y;
  • double z = feature_msg->points[i].z;
  • // （p_u, p_v）：Pixel coordinate
  • double p_u = feature_msg->channels[2].values[i];
  • double p_v = feature_msg->channels[3].values[i];
  • // The motion speed of feature points on the pixel plane
  • double velocity_x = feature_msg->channels[4].values[i];
  • double velocity_y = feature_msg->channels[5].values[i];

[shiba24]

@SharineLee Thank you for the idea, I like it. First of all, let me clarify this is a dense optical flow estimation method. So if you meant by "feature_tracker" as the velocity (2-dimensional displacement speed) for some sparse points (pixels) in the image plane, the dense optical flow estimation might be "overkill" - too expensive.

Yet, if you want to try to integrate this module into your ROS node, I'm happy to help you. That was on my personal todo list but I was too busy to do that. Also, please see my replies for your other issues if this code really matches your use-case. #11 #12

Best