KUKA Delta Robots PTP vs LIN

Quote from jaiiyer

Hi,

For Delta robot does anyone use PTP motions while using with a conveyor? PTP motions results in collision with the conveyor if the pick height is very small. However PTP motions are way faster than LIN motions. (about 10% faster as per the video)

I know about the hop function but it hasn't been working well for me on KUKA SIM and don't when if I need the Arctech license for it to work on KRC5.

Why cant the PTP motions be in a way that the arc is away from the conveyor? I don't have much background on robot kinematics and if that's the reason then understood and I am sorry for the question.

My thinking is that most of the delta applications are with conveyors and having the delta is restricted in its ability to have fastest motion seems like a headache.

Did you tried and put additional middle PTP move in place where delta hits the conveyor so that it doesn't hit the conveyor anymore?

Quote from jaiiyer

yes but that increases the time taken to finish the move and for vision with conveyor tracking it will be difficult to calculate the intermediate point. Not that it can't be done.

In my cases with PA robots where points were below robot's base and the robot wanted to collide with the conveyor, it didn't slow down the robot much by adding the intermediate PTP point, but I usually don't have points very close together with PA robots, and with using correct load data and inertia managed to squeze speed out of the robot. Also LIN moves can be tweaked with ori and acc parameters so much that the controller can't hold the robot anymore on the programmed path when going beyond the robots limits.

You are trying to cheat physics. You have a linear conveyor, so you have to move linear.

When you want to move ptp not hitting the conveyor, you have to calculate intermediate positions above the conveyor. But the position should not be too much above the conveyor. After all, when you have done your own calculations for the intermediate positions you get nearly a linear movement, and the same limitations like standard linear interpolation. That's because during the complete movement some of the axis have to accelerate and deccelerate more than once. On ptp movements each axis only accelerates and deccelerates once.

As Herman says you can not cheat physics. PTPs are LINs in axis space. So they connect given axis positions by a straight line in axes space. This results only implicitly in the cartesian move you see. Your expectation is to specify a cartesian path to follow by ptp. Thats nonsense and not possible. You can not follow a straight line with your axes and at the same time a different cartesian path. In LINs it is the other way round you specify a cartesian path and the axes follow accordingly. By dividing the path into smaller PTPs you essentially approximate a cartesian path yourself and encounter the same problem as with directly using LINs. Your move becomes slower. In newer controller versions AFAIK the system now also uses a dynamic model behind the scene to calculate the time optimal trajectory for LINs. Som3thing you get when using SLINs instead of LINs. But AFAIK Spline does not support conveyor yet. But there is nowadays a limited dynamic model planning for LINs as well. Not sure if this is now available all the time with LINs on conveyor or only in some special cases like hops in ArcTech. A question you could talk about with your KUKA representative.

Fubini

Para utilizar un robot Delta en una cinta transportadora y realice movimientos lineales a maxima velocidad se utiliza el software ConveyorTech, este tiene una funcion de movimiento bloque DLIN, alli se programan movimiento Dlin (lineale) estos se ejecutan a su maxima velocidad que puede realizar el robot son mucho mas rapidos que los movimiento lineales comunes LIN.