## emc-developers

 [Emc-developers] Traj calculations From: andy pugh - 2013-08-15 11:56:06 ```I seem to recall discussion of using the kinematics module to perform a discrete differentiation to determine the trajectory speed. (which would allow velocity limits to be respected in non-trivial multi-axis kins). Is this implemented anywhere? -- atp If you can't fix it, you don't own it. http://www.ifixit.com/Manifesto ```
 Re: [Emc-developers] Traj calculations From: Jeff Epler - 2013-08-19 16:08:55 ```On Thu, Aug 15, 2013 at 12:55:38PM +0100, andy pugh wrote: > I seem to recall discussion of using the kinematics module to perform > a discrete differentiation to determine the trajectory speed. (which > would allow velocity limits to be respected in non-trivial multi-axis > kins). Is this implemented anywhere? I think this is the Inverse Jacobian, and is implemented in genhexkins but unused by linuxcnc. http://en.wikipedia.org/wiki/Jacobian_matrix_and_determinant The Jacobian at a point is not itself enough to figure out how to apply joint velocity constraints, because those constraints vary all along the course of the move. A single move, even a linear one, can take the machine from a place where joint J0 would require a velocity limit, into a region where the F-number gives the velocity limit, and then into a region where joint J1 gives the tighest velocity constraint (crossing the center of the work volume of a delta, for instance). However, computing the jacobian at many points along a planned move may be one ingredient of applying joint velocity constraints to trajectory planning. Jeff ```
 Re: [Emc-developers] Traj calculations From: andy pugh - 2013-08-19 16:24:29 ```On 19 August 2013 16:52, Jeff Epler wrote: > The Jacobian at a point is not itself enough to figure out how to apply > joint velocity constraints, because those constraints vary all along the > course of the move. A single move, even a linear one, can take the > machine from a place where joint J0 would require a velocity limit, into > a region where the F-number gives the velocity limit, and then into a > region where joint J1 gives the tighest velocity constraint This seems to imply that for a linear move the kins transformation is only applied at the end points, and that was not my understanding. Or is the assumption that the velocity limitation has to be on a move-by-move basis not a point-by-point basis? Does adaptive-feed take effect mid-move? -- atp If you can't fix it, you don't own it. http://www.ifixit.com/Manifesto ```
 Re: [Emc-developers] Traj calculations From: Michael Haberler - 2013-08-19 16:27:19 ```see this project https://trac.lcsr.jhu.edu/cisst , it might be a cheat sheet in particular references here: https://trac.lcsr.jhu.edu/cisst/wiki/cisstRobotTutorial -m Am 19.08.2013 um 18:23 schrieb andy pugh : > On 19 August 2013 16:52, Jeff Epler wrote: > >> The Jacobian at a point is not itself enough to figure out how to apply >> joint velocity constraints, because those constraints vary all along the >> course of the move. A single move, even a linear one, can take the >> machine from a place where joint J0 would require a velocity limit, into >> a region where the F-number gives the velocity limit, and then into a >> region where joint J1 gives the tighest velocity constraint > > This seems to imply that for a linear move the kins transformation is > only applied at the end points, and that was not my understanding. > > Or is the assumption that the velocity limitation has to be on a > move-by-move basis not a point-by-point basis? > > Does adaptive-feed take effect mid-move? > > -- > atp > If you can't fix it, you don't own it. > http://www.ifixit.com/Manifesto > > ------------------------------------------------------------------------------ > Introducing Performance Central, a new site from SourceForge and > AppDynamics. Performance Central is your source for news, insights, > analysis and resources for efficient Application Performance Management. > Visit us today! > http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk > _______________________________________________ > Emc-developers mailing list > Emc-developers@... > https://lists.sourceforge.net/lists/listinfo/emc-developers ```