nadel@aerospace.aero.org (Miriam H. Nadel) (07/04/90)
In article <11485@rasp.eng.cam.ac.uk> jjs@eng.cam.ac.uk (Jeffrey J. Shifman) writes: >I am looking for either > >(1) suggestions for how to simulate a flexible robot (say, for simplicity, > two flexible links modelled as Euler-Bernoulli beams, with a single > torque input to each one) with a peculiar feedback law; or, > >(2) a simulator that can deal with the system described above. > >I have tried a 'naive' modal approach but it doesn't seem to work. >Any suggestions appreciated. > There are three basic approaches to modeling flexible robots. The modal approach is pretty common but the big drawback is the inaccuracy introduced by the need for modal truncation. In addition, it's not simple to deal with mechanisms. That is, the modal approach doesn't combine easily with a model of, say, a motor. The second approach is to discretize the flexible links, e.g. via a finite element model. This handles mechanisms much better but is very intensive computationally if you want to get halfway decent result. It also ignores the effect that the elastic motion of the beams has on the rigid body motion of the joints. The third approach uses generalized coordinates to derive the equations of motion and then incorporates the joints using Lagrange multipliers. This is the only approach which allows both the rigid body motion and elastic motion to be influencing each other. I would suggest looking up the two articles by Nagarajan and Turcic in the June 1990 issue of the ASME Journal of Dynamic Systems, Measurement and Control (which is also a generally good source of info on robotics and modeling). Miriam Nadel -- One of the 20% of Americans who own stock. nadel@aerospace.aero.org