Dynamic Manipulability
|
Dynamic Manipulability | |
|
| Dynamic manipulability refers to a manipulator's capacity to accelerate its end-effector in various directions. This differs from the usual measures of "kinematic" manipulability, which quantify a manipulator's velocity and force transmission capabilities with no explicit role for the actuators or the dynamics. As part of this project I derived the effects of velocity on dynamic manipulability, and this provided the theoretical basis to show that certain solutions to the robot weightlifting task are fundamentally better than others. Measures of dynamic manipulability are useful not only for analysis of observed movement, but also as a source of prior knowledge with which to seed a learning problem or trajectory planner. Interestingly, a coarse-grained measure of the velocity-dependent effects can be had from manipulator positions alone, i.e., without prior knowledge of velocity. |
|
Related Publications | |
| Velocity-dependent dynamic manipulability | |
| M.T. Rosenstein and R.A. Grupen. In Proceedings of the IEEE International Conference on Robotics and Automation, vol. 3, 2424-2429, 2002. [pdf] [ps.gz] | |
| Robot weightlifting by direct policy search | |
| M.T. Rosenstein and A.G. Barto. In Proceedings of the Seventeenth International Joint Conference on Artificial Intelligence, vol. 2, 839-844, 2001. [pdf] [ps.gz] | |
|
Movies | ||
| File | Description | Download |
| vel_bias | A slightly confusing animation that illustrates the effects of velocity on manipulability. (The animation is confusing because the manipulator configuration remains fixed while the velocity at that configuration increases as the animation progresses.) Shown in blue is the kinematic velocity ellipse and shown in green is the dynamic manipulability ellipse corrected for gravity. The red polytope shows the robot's true acceleration capabilities and accounts for inertia, gravity, and velocity. Velocity has no effect on the polytope shape yet introduces a bias that shifts the center of the polytope away from the end-effector. | 820 KB |
| manip_slomo | Slow-motion animation of one solution to the weightlifting task. The small red polytope depicts the acceleration capabilities of the end-effector. Notice that the polytope is quite small for this example yet the velocity bias, i.e., the displacement of the polytope, can grow quite large. | 955 KB |
| manip_zoom | Close-up of the acceleration polytope from the preceding animation. The red dot shows the acceleration command chosen by the control system throughout the movement. Accelerations at the edges of the polytope correspond to torque limits of at least one joint actuator. | 2.5 MB |
| Note: To view the movies, use a QuickTime player (Mac/Win), MPlayer (Mac/ Linux), or at least version 2.80 of XAnim (Unix). | ||
|
updated 29-Nov-2003 mtr@cs.umass.edu |