BALTIMORE Engineers at Johns Hopkins University have implemented a spherical motor that can roll in any direction, rather than rotate on a single axis as conventional motors do. Spherical motors could enable omnidirectional "wheels" on everything from robots and forklifts to computer mice, developers say.
Robotic arms that use the spherical motor would work like ball-in-socket joints and would require a fraction of the number of motors needed for today's elbow-like robotic joints.
Spherical motors are not new and have been researched elsewhere. The Georgia Institute of Technology (Atlanta) has demonstrated an induction-based spherical motor suitable for free-running applications, for example.
"We didn't invent the idea of a spherical motor, but we have added the engineering refinements to implement the first spherical stepper motor that could replace conventional motors for many applications," said professor Gregory Chirikjian, who performed the work with research assistant David Stein and professor Edward Scheinerman. The researchers have applied for two U.S. patents on the spherical motor, which Chirikjian and Stein detailed at a recent meeting of the American Society of Mechanical Engineers.
The Johns Hopkins motor surrounds its core called the ball with a spherical shell studded with electromagnets controlled by a computer. By activating the proper electromagnet, magnetic force can be applied to cause the permanent magnets inside the ball to roll it in any desired direction, one step at a time.
In the prototype, 80 permanent magnets were mounted inside the ball and made to respond to 16 electromagnets encircling the ball. In practice, two of the electromagnets must be activated to roll the ball to a new position.
Robotic arms today depend on six or more conventional motors to position and orient their actuators in three dimensions. That's because the arms must be fitted with multiple internal joints that only bend in the one direction of each motor. The software driving those multiple motors must then solve numerous transcendental equations to translate 3-D positions in space into the linear motions of the six internal axes.
Spherical motors, however, would enable the internal joint of the robot's arm to be a true ball-in-socket rather than just emulate the ball-in-socket using multiple independent axes. As a result, they would require less internal hardware, and the software would need to solve fewer transcendental equations, instead executing quick linear programming to accomplish the same tasks.
"With a single spherical motor, you get three degrees of freedom, so theoretically you could replace three conventional motors on today's robotic arms with just one spherical motor," said Chirikjian. "However, when we connect two spherical motors together with a solid rod, one degree of freedom becomes internal, so you only get five degrees of freedom with two, and seven degrees of freedom with three. So we say it will take three spherical motors to replace a six-degree-of-freedom, conventional robotic arm."
Fewer parts, higher accuracy
The current phase of the project, Chirikjian said, is "experimenting with the motor itself, not in an application, such as for a robotic arm. But if you did build a robotic arm with our spherical motors, it would have far more freedom of movement than a conventional robotic joint, and each joint would have much less jiggle because of inherently higher accuracy and fewer moving parts."
The encoder for the spherical motor works by observing black and white stripes on the ball, with 96 pairs of photo-emitters and photodetectors mounted in a hexagonal pattern around each of the 16 electromagnets in the shell. Also mounted in the shell are casters that bear the weight of the ball as it rotates, enabling it to handle any conceivable load by merely beefing up the casters appropriately.
The prototype uses a sealed shell, but by allowing a bit of the ball to protrude from the shell much as a computer mouse's ball protrudes from the bottom of the mouse the spherical motor can drive external devices or act as a wheel for vehicles.
For instance, omnidirectional spherical tires could be turned in any direction to allow parallel parking in extremely tight spaces or other precision maneuvers.
Researchers also plan to experiment using spherical motors to replace the spherical ball on the bottom of computer mice. Force feedback could thus be issued to guide the hand of the user in automated help systems, for example, or to enhance simulations.
"We can imagine all kinds of utilitarian uses for allowing intelligent agents to take control of your mouse," said Stein. "It would add a whole new dimension of tactile interaction, since today computers depend almost exclusively on visual feedback. For 3-D simulations, it could simulate bumping into solid objects."