I've added a capture method for the CMVision driven blobfinder device that
supports the Player "camera" interface. This is neater than using the removed
captureGazebo and the other gz_camera.cc for finding blobs in Gazebo images.
It's also an alternative to using capture "V4L".
I made a rudimentary effort to support the camera interface geometry changing
online. CMVision is reinitialized when it happens. Only did this because it
can happen. Probably best to avoid it. Both Gazebo and Player can crash when
Example of player configuration:
On Sunday 11 January 2004 13:26, ahoward wrote:
> Hi folks:
> I've initiated a major overhaul of camera support in both Player and
> Gazebo (nothing like a DARPA demo to get those fingers coding). The
> changes are as follows...
> - The Player "camera" interface has been fleshed out. It now contains
> information about image size and depth, together with the actual image
> data (as an array of uint8_t). Thus, the player_camera_data_t structure
> can be used to pass around entire images. Note, however, that this
> structure is *too big* to be passed over the network. Instead, the camera
> interface is used *internally* within Player, to pass data around between
> drivers (more on this below).
> - Currently, there are two drivers supporting the camera interface:
> gz_camera and camerav4l. The first captures simulated images generated by
> Gazebo (OpenGL-based rendering), the second captures images using the
> Video for Linux (version 1) API.
> - I have created a new fiducial detector called laservisualbw that detects
> combined laser/visual barcodes. This driver subscribes to both laser and
> camera devices (detecting retro-reflectors from the first and UPC barcodes
> from the second). Importantly, since the underlying devices are
> abstracted through their respective interfaces, this driver works equally
> well with the gz_laser and gz_camera drivers as it does with the
> sicklms200 and camera4vl drivers. I was, in fact, able to do almost all
> of the development of this driver in simulation.
> - The existing CMVision hacks are unchanged, but hopefully someone will
> see fit to migrate them over to this new model.
> - In Gazebo, the SonyVID30 model has been upgraded to support full
> pan, tilt and zoom functionality, and return images via the camera
> interface. When Nate's new render engine is ready, we should be able to
> do full texture maps, fog and so on.
> Andrew Howard email: ahoward@...
> Department of Computer Science http:
> www-robotics.usc.edu/~ahoward University of Southern California
> phone: 1 (213) 740 6416
> Los Angeles, CA, U.S.A. 90089-0781 fax: 1 (213) 821 5696
> << Insert pithy saying here >>>
> This SF.net email is sponsored by: Perforce Software.
> Perforce is the Fast Software Configuration Management System offering
> advanced branching capabilities and atomic changes on 50+ platforms.
> Free Eval! http://www.perforce.com/perforce/loadprog.html
> Playerstage-developers mailing list