Hi Everyone,
I am currently trying to solve an issue I am having with path recording and playing back the recorded path.
As of this moment I am able to:
1. Record the robot joint angles of the LBR iiwa 7.
2. Get the frames at each recorded position.
3. Use the frames to create a SPL motion and have the robot run it.
The issue with the method above is that the motion is generally not smooth and has very unreliable movements.
We believe that our current method is using too many points that are close to each other
An idea between myself and my colleagues is that we create a Interpolated 3D Spline for these points and then use the formula of the created spline to create equidistant points along the curve. This would allow use to control the smoothness of the planned path. So the proposed plan of attack is now:
1. Take recorded poses (robot joint angles).
2. Run poses through forward kinematic chain with redundancy to get x,y,z,a,b,c,r
3. Fit 3D Spline to those points with DIRECTION!!!
4. Use a distance delta to get new equally distance points.
5. Take new x,y,z,a,b,c,r positions.
6. Run it through kinematic chain to get final poses.
I am sure I will be able to figure out 3 and 4 with time, but my main issue is how to get a reliable r value from step 4 to 5.
Any thoughts or points to the right direction would be helpful.
Thank you!