Each car has a ZED stereo camera. It can output a standard video stream, or it can output a depth estimate stream.
- Sends data and receives power via USB
- Frames per second & resolution: 60fps @ 720p
- Field of view: 110°
- Depth range: 0.5 - 20 m
- Can calculate odometry based on visual data
This sensor has an MSRP of 0.5 US kilodollars. Be aware of this as you conduct your testing.
Usage in ROS
- To use the ZED's data in ROS, first run `startZED` on the car's terminal.
- Beyond that, you usually will not have to interact with ZED data directly. Rather, we will often give you programs that process the ZED’s data.
For instance, in ZED.py , the “image_converter” class takes in the raw data and converts it to CV2 format, which helps us greatly to process the image.
For the curious though, we can get ZED data like we get other sensor data
- The ZED nodes publish different data to different topics. There are left/right camera topics and a depth topic. We often use `/zed/rgb/image_rect_color` .
- Each message is of the type `Image`.
- For a full list of `Image` message attributes, see ros.org.
How to visualize the data onscreen.
- If you're ssh-ed into the car:
- on the car's terminal (i.e. ssh-in), run `startZED`
- on the computer's terminal, run `rqt_image_view`
- select the topic you would like to view
- If you have a monitor directly plugged in:
- Run `startZEDviz` in the terminal for a totally awesome visualization in RVIZ.
- Be sure to click on "DepthCloud" for the full effect.