The forward kinematics solver does solve for the link positions. If you take a look at the service definition, http://www.ros.org/doc/api/kinematics_msgs/html/srv/GetPositionFK.html it takes in an array of link names and a joint state, and outputs an array of poses for the links you specified. Thinking about it more, since the origin of the link is defined to be the origin for the joint, I could see how you could see the FK as solving for the joint "poses". If that doesn't help, can you send out your URDF and the data you are sending to the get_fk service? -David!! On Sat, Dec 4, 2010 at 10:19 PM, David Lu!! wrote: > I thought it was for the Link, and will check for sure when I get back to > my office tomorrow or Monday. > > -David!! > > > On Sat, Dec 4, 2010 at 8:12 AM, Patrick Goebel > wrote: > >> OK, I think I have answered my own question and sorry for making a mess >> of the original question. The question should have simply been: does >> the *forward* kinematics solver map joint positions to poses of the >> joints or the links? Since the origins of joints and links need not >> coincide, I thought it could be one or the other. >> >> Further experimentation seems to prove that the poses computed by get_fk >> are for the joints, not the links, although both the documentation and >> the output from get_fk use the word "Link" to label the computed poses. >> >> Do I have this correct? >> >> Thanks! >> patrick >> >> _______________________________________________ >> ros-users mailing list >> ros-users@code.ros.org >> https://code.ros.org/mailman/listinfo/ros-users >> > >