Reorienting with small radius with Force Control

xerim

Hello,

Has anyone ever had any success with large reorientations of the TCP to follow a small radius with force control enabled? I've been trying to figure this out for what seems like ages now. I've been experimenting with custom zonedata and speeddata to control the size of the zone necessary for the next reorientation to begin but with no luck. It seems that at the apex of the curve the tool always pushes far too hard, decreasing the force doesn't help this. I've tried moving the target further away yet it consistently drives the tool further than I would like when trying to navigate to the next target. My theory is that in the process of reorienting to achieve the next position, the force control is still searching for the direction in which to apply force, causing it to apply an unnecessary amount of force when making many reorientations in a short period of time (FC_REFFRAME is set to tool workspace here). The way I have it programmed now is with many, many points in order to increase the smoothness of the motion. I have found that the problem gets worse with less points. Can anyone shed some light on this problem? What am I missing here? 

xerim

Here is a more visual description of the problem. The tools drawn in green perform as expected, while the tools drawn in red represent orientations that give undesirable results (too much pressure). I have tried creating my own zonedata that allows for smoother reorientation as well as my own speeddata to control the reorientation speed but nothing seems to work, by the time the TCP reaches the first tool drawn in red the pressure always extends vastly beyond what I have programmed it for. 

lemster68

Sorry if this comes as an insult, but I have to ask...how good is your TCP?

xerim

I tested the TCP by reorienting the tool in the same orientations depicted above and noticed little to no drift at all, I suppose I could re teach the TCP but I don't think this is the issue as I have no problems doing similar orientations in other programs. The only difference is the programs with more extreme reorientations don't implement force control, only linear movements. Force control seems to lose it's accuracy when reorienting the tool, even at low speeds. I suppose its just a matter of trial and error..

lemster68

OK, next question, where is the TCP?

xerim

The tool is a 254mm diameter circular buffing wheel, and the TCP is at the apex of the circle at the middle point of the thickness of the tool. 

lemster68

Well, that sounds about right, I was wondering if it was in the center of the wheel.

xerim

Ah no, I wish it was a simple fix like that though. I am going to continue experimenting with FCSetDampingTune, zonedata, speeddata and the robtargets I've taught. If I have some major breakthrough I'll be sure to post the solution in here.