|
Re: Image Processing
There are a couple well supported ways to get the images to the dashboard. The initial one was necessary because there was no switch on the robot. This meant that the cRIO asked the camera for the images on one enet port, and then duplicated the jpeg out the other port to the dashboard.
For a few years now, the bridge has included a switch. This means that the dashboard and camera can directly communicate. This is usually done by having the dashboard request an mjpg stream from the camera. The camera can serve up to five clients, so it is actually possible to have images go to the cRIO and dashboard independently.
To get processed results back to the robot from the dashboard is not quite as automatic. One option is to use the Network Tables -- the underlying publish-and-subscribe protocol that SmartDashboard uses. This was relatively new last year, but is being enhanced and is what I'd recommend. It is also possible to do your own lower level protocol over TCP or UDP. Network tables operates over TCP by the way.
As for the speed of the connection, the robots use N-speed wifi, a high speed communication option. It will typically take just a few milliseconds for the information to be sent, and in most measurements we've seen, the robots are using less than 25% of the bandwidth available.
Greg McKaskle
|