I recently had the problem of having a 2-axis system that I needed to integrate in the transformer. I ended up doing "the standard thing", that is hard-coding the device's geometry in the driver component and generate the equivalent RBS internally. The annoying bit about that was the inability to test the visualization without either having the device or making a mock transform generator that matched the driver's transformation generation (with the usual issue of how to keep both synchronized).
So, I thought, what about we start using all these model files ? The general idea I had was to extend the transformer to be able to feed joint data instead of transform data. It would require adding a robot_model property and a dynamic_transformation_joints port. The first would be the path to a model file (URDF or SDF), which would internally be converted to a KDL structure. The static parts would be stored as static transforms, the dynamic parts fed when the relevant joint states are received. Thoughts ? Sylvain _______________________________________________ Rock-dev mailing list [email protected] http://www.dfki.de/mailman/cgi-bin/listinfo/rock-dev
