-
Notifications
You must be signed in to change notification settings - Fork 1
Example: Simulated Grating
- one KUKA LWR
- simulator based on RobotToolKit from EPFL
- three virutal objects: a grater, a trash bin, and a supply box
The robot shall perform grating of virtual carots in a simulated environment. For that it needs to pick the carots up at a supply box, move to the grater, perform several grating motions, and finally drop the carot stop over the trash bin.
In this example, we fill focus on software integration and not on realistic physics simulation. To further simplify, we first only use the CRAM executive and the SEDS controllers to achieve this task. In a second step, we will relax some of the hard-coded decisions in CRAM by using knowledge processing in KnowRob.
Since we use an environment without accurate task simulation, most of the task execution errors which one would typically see in the real-world are missing. Hence, the task executive will be smaller and will mainly deal with selecting the right tasks to execute. Still, we hope that the concepts behind CRAM and SEDS become clear in this example, and that some integration issues surface early.
- CRAM as task executive.
- SEDS (or CDS ?) to perform grating motion and Cartesian movement to pick up and throw away carots.
- PID controller to execute joint motions (does RobotToolKit have this?)
- 'fake' perception module providing the poses of the three objects in the world
- robot simulator from EPFL
Components 1.-4. communicate through ROS. Components 2.-4. communicate to 5. through native C++ API calls.
TODO: add diagram for software components.
TODO: break the software components into smaller parts to explain their workings..