ZED ROS2 Wrapper Outputs

In this article, there will be short descriptions/references for currently relevant outputs from the ROS2 ZED Wrapper node that are pertinent to this project. This article will be continuously updated to include every new output from the wrapper that is being utilized by the project. Let somebody from the perception group know what new outputs you want if it is not listed down below. All available topics are listed in the official ZED docs, in this article, under “Published Topics”.

Object Detection

Topics: ~/obj_det/objects: objects detected frame by frame

The topic above is an array of zed_interfaces/Object messages, which is I will describe:

• string label: this is the object’s label, in other words the class type. Ex: “PERSON” or “VEHICLE”
• string sublabel: the object’s sublabel, which is a class type falls under the label category but is more specific.
  • Ex: if the object is a bicycle, then label = “VEHICLE”, and sublabel = “BICYCLE”
  • Comprehensive list of all object classes and subclasses are listed here.
• float32 confidence: the confidence level of the object detector is essentially the probability that the detector thinks the object exists within the bounding box.
  • values can range from 1-99
  • worse values mean the detector is having hard time localizing the object or that it is not sure about the label.
• float32[3] position: this float array describes the centroid of the detected object, relative to the world frame. Can be interpreted as a position vector (px, py, pz).
• float32[3] velocity: this float array describes the velocity of the detective object, relative to the world frame. Can be interpreted as a velocity vector (vx, vy, vz).
• zed_interfaces/BoundingBox3D bounding_box_3d: this field represents the 3 dimensional bounding box that outlines the object, relative to the world frame.
  • Contains the field zed_interfaces/Keypoint3D[8] corners, which is an array of 8 float32[3] kp objects, each of which represent a position vector (px, py, pz).
  • The indices of corners can be visualized as:

       1 ------- 2
      /.        /|
     0 ------- 3 |
     | .       | |
     | 5.......| 6
     |.        |/
     4 ------- 7