IEEE Robotics & Automation Magazine - December 2010 - 14

E N T R E P R E N E U R

Butterfly Haptics: A High-Tech Start-Up
Ralph L. Hollis, Butterfly Haptics, LLC, http://butterflyhaptics.com

here are many ways to contribute to the exciting and
growing field of robotics. We all try to publish good
papers and participate in conferences. A few of us also
see a way to contribute by offering state-of-the-art products,
which can be used as platforms for continuing research by the
community. Our company, Butterfly Haptics, LLC, is one
such company. We produce and sell high-performance haptic
systems based on a magnetic levitation principle. The systems
can be interfaced to virtual environments and remote real environments to provide high-fidelity touch interaction.

The Technology
Researchers have experimented with haptic devices for several
decades. Fred Brooks and students at the University of North
Carolina were early pioneers. Ken Salisbury and his student at
the Massachusetts Institute of Technology (MIT) developed a
haptic device that was commercialized as the Sensable Technologies PHANToM. This and its follow-up products are
likely the most widely used haptic devices. In recent years,
half-dozen companies have jumped into the haptic arena. All
of these companies' devices are essentially lightweight backdriven robot arms, with either serial or parallel kinematics,
ranging from two to six degrees of freedom (DOFs). Because
they are robot mechanisms, they have links, joints, transmission
elements, motors, and encoders. These elements are subject to
wear and tear, link flexing, cable stretching, joint friction, backlash, and motor cogging effects, which together can interfere
with the fidelity of haptic interaction.

2
3

4
5

Figure 1. Cutaway view of the Butterfly Haptics Maglev 200
device: 1) handle, 2) levitated flotor, 3) magnet assemblies, 4)
LED beacons, 5) optical sensors, 6) flexible wires, and 7) interface.
Digital Object Identifier 10.1109/MRA.2010.938835

IEEE Robotics & Automation Magazine

We have taken a different approach based on principles of
magnetic levitation. The user interacts with a handle rigidly
connected to a hemispherical structure called a flotor that is
freely levitated in magnetic fields. The handle with its flotor
can be translated and rotated by the user in six DOFs within
the work space. Optical sensors measure the position and orientation of the handle, providing input to the user's computer,
while six coils embedded in the flotor interact with the fields
of stationary permanent magnets through the Lorentz force to
provide a force-torque wrench at the handle. The result is
more than an order of magnitude improvement in performance compared with current electromechanical haptic devices.

How We Got There
Magnetic levitation haptic devices (Figure 1) seem pretty
straightforward to us now, but looking back at their origins
some 26 years ago, it was not quite so obvious. I was a researcher
at the IBM Thomas J. Watson Research Center and had invented
several two- and three-DOF robotic fine motion devices for use
in automated assembly and testing. I wanted to create a full sixDOF device with reasonable motion range but was baffled by
how to design the appropriate linkages. It occurred to me one
evening that by combining recently introduced high-strength
magnetic materials with position sensing optical diodes and the latest digital signal processing, one could eliminate all the mechanics
and precisely control a freely floating body having but a single
moving part.
Several years later with help from Peter Allan, a Duke
University summer student intern, and controller design by
Tim Salcudean, a new Ph.D. hire from Berkeley, we had a
crude version working on the bench top using a collection of
surplus power amplifiers and a US$9,000 digital signal processor plugged into an IBM PC/XT. In subsequent years, a
refined version that could be attached to the terminal link of a
robot was built and replicated as the IBM Magic Wrist. We
were able to synthesize a number of compliant mechanisms
for parts insertions and demonstrate its viability in coarse-fine
robotic applications. We also interfaced it to an early scanning
tunneling microscope to enable feeling of atomic-scale surface
features in real time.
We soon began to realize that the operating principle of
our devices, which we dubbed Lorentz levitation, could be
applied to several other domains, including vibration isolation
(particularly in space), haptic interaction with virtual environments, and teleoperation. Tim moved on to the University of
British Columbia, where he continued to work with Lorentz
levitation, and a few years later, I joined Carnegie Mellon
University in Pittsburgh. Tim and his students built a system
DECEMBER 2010

http://www.butterflyhaptics.com

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2010