Volume 19 Issue 07 - Publications Date: 1 July 2000
Lorentz Magnetic Levitation for Haptic Interaction: Device Design, Performance, and Integration with Physical Simulations
P.J. Berkelman The Robotics Institute, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, USA and R.L. Hollis
A new Lorentz force magnetic levitation haptic interface device has been developed and integrated with real-time 3-D rigid-body simulations for detailed, responsive interaction with dynamic virtual environments. An ideal haptic interface device would enable remote or simulated objects to be sensed and manipulated in 6 degrees of freedom in the same manner as real physical objects, with the same micrometer level of detail and kHz bandwidth response that the user can sense. To approach this level of performance requires stiff lightweight moving parts, responsive actuators, high-resolution sensors, a fast control system, and a real-time simulation closely integrated with the device. Lorentz levitation is well suited to high-performance, tool-based haptic interaction because noncontact actuation and sensing provide motion and force feedback in 6 DoF with the simple dynamics of a single moving part at high control bandwidths and sensitivity. New actuation and sensing subsystem designs greatly increase the user range of motion over previous maglev devices. The motion range of the new device is ±12.5 mm in translation and ±7.5 deg in rotation to accommodate haptic fingertip motion tasks. The device has a measured closed-loop position bandwidth of over 100 Hz in each DoF, a maximum stiffness of 25.0 N/mm, and a position resolution of 5-10 µm. Virtual coupling and intermediate representation methods were implemented to combine the simulation and the device controller and maximize the realism of haptic user interaction when computational resources are limited. In our system, the device controller must cycle several times faster than the simulation to generate stiff constraints during stable levitation. With virtual coupling, the updated position and orientation data from the interface device and the simulated tool act as error and velocity feedback control setpoints for each other. With the contact point intermediate representation method, a list of current tool contact points and directions is generated at each update of the simulation so that the device controller can evaluate contact constraints locally and change impedance to respond to user motions at the fast control rate rather than the slower simulation rate. The contact point intermediate representation provides a crisper, more responsive feeling than virtual coupling during interaction but is not as easily stabilized. Experimental data from the virtual coupling and a modified contact point intermediate representation are presented.
Return to Contents