MODERN manufacturing makes heavy use of robots, which are better than humans at repeating the same task over and over. But when even minor changes need to be made in the manufacturing process--in the shape of a car door, for example--a human operator must "teach" the robot the new shape by guiding it by hand through each motion and every orientation in the operation. Besides being time consuming and therefore expensive, this process is often inaccurate.
Charles Vann, a Lawrence Livermore National Laboratory mechanical engineer and manager, has developed a sensor that can make manufacturing robots smarter, saving both time and money. His small, noncontact, optical sensor increases the capability and flexibility of computer-controlled machines by detecting the sensor's relative position to any mechanical part in all six degrees of freedom. (In mechanics, degrees of freedom refer to any of the independent ways that a body or system can undergo motion, i.e., straight-line motion in any one of the three orthogonal directions of space or a rotation around any of those lines.) The six-degrees-of-freedom (SixDOF) sensor can be mounted on the tool head of a multi-axis robot manipulator to track reflective reference points attached to the part. Once the robot knows where it is relative to the part, a computer can instruct the robot to follow a path predescribed in multidimensional computer drawings of the part, or the robot can be programmed to follow a path of references mounted on the part. The sensor eliminates the need for "training" the robot and enables process changes without halting production because software can be downloaded quickly into the robot's controller.
The nearest competitor to the SixDOF sensor is one that detects only three degrees of freedom. But many manufacturing operations require information on all six degrees of freedom. Welding, for example, requires information on three degrees of freedom to locate the weld (the x, y, and z axes) and the other three rotational degrees of freedom to properly orient the tool relative to the part. Compared to the competitor, the new SixDOF sensor is four times smaller and five times lighter because it uses lateral-effect photo diodes (light- and position-sensitive diodes), which are smaller and lighter than the cameras used by the competition. And the SixDOF sensor costs one-sixth as much. Yet for an equivalent field of view, it is more than 250 times faster and up to 25 times more accurate.

How the Sensor Works
The SixDOF sensor is composed of four assemblies: a laser illuminator, beam splitting and directing optics, lateral-effect photo diodes, and signal-processing electronics. The laser source is a 5-milliwatt diode laser. Two small mirrors (M1 and M2 on the illustration, next page) guide the 1-millimeter laser beam to the primary optical axis of the sensor. The beam then passes through two negative lenses (L1 and L2) that diverge the beam at about 0.3 radians. This high divergence creates a 2-centimeter laser spot at about 3.5 cm from the face of the sensor. The beam divergence, depth of field, and spot size can be changed by choosing different negative lenses.
Two reflective reference points, a 4-millimeter dot and a 1-by-1-mm bar, are mounted on nonreflective tape and applied to the part being worked on. The laser light reflects off the references and back into the sensor. Because the beam is diverging, the reflections are magnified in area when the light returns to the sensor, allowing most of the light to go around the negative lenses and through a large, collimating lens (L3) instead. After collimation, the beam continues through a notch filter, which passes the laser light but blocks light at other wavelengths.
Inside the sensor, light from the dot is divided into two beams by a beam splitter. Half of the beam is reflected 90 degrees into photo diode P3. The other half of the beam passes through the beam splitter, into a focusing lens L3, and onto photo diode P2.
The light from the second reflective surface, the bar, also passes through the filter. However, because this reflective bar is tilted relative to the dot, the laser light reflecting from it is at a greater angle of divergence. The greater angle causes the light to pass through a different location of the filter, missing the collimating lens and illuminating another photo diode (P1).
Through creative use of mirrors and lenses, each of the three photo diodes has a different sensitivity to the relative position of the sensor and the reflectors. P1 is most sensitive to straight-line motion between the bar and the sensor z and the rotation of the sensor about that axis (Rz). P2 is most sensitive to tilt about the x and y axes (Rx and Ry), and P3 is most sensitive to straight-line motion of the sensor relative to the reference dot (x and y). Information from all three sensors is needed to determine all three positions and three orientations of the sensor relative to the part.
The signals from the three photo diodes are processed by electronics remotely located from the sensor head. The analog data from the diodes are digitized and fed into a computer where they are decoupled to define the six axes of information. The processed data are then available to the operator for recording or sending commands to change the position of a computer-controlled machine.

A Better Mouse
Among other future uses for Vann's new sensor is a SixDOF cursor for personal computers, which would allow a user to perform much more complicated tasks than are possible today with a typical two degrees-of-freedom mouse. The sensor could also be used to help doctors diagnose muscle recovery by evaluating the effects of physical therapy. With reflective reference points mounted on a patient's injured limb, a robot with a SixDOF sensor could generate a SixDOF map of muscle motions. The sensor could also remotely perform dangerous tasks such as manipulating radioactive, toxic, or explosive materials. For example, a robot with a SixDOF sensor could track reflective references mounted on the hands of an operator who disassembles a dummy bomb while another robot, electronically following the motions of the first robot, disassembles the real one.
Its potential applications are diverse, but the SixDOF sensor will likely find its greatest use in manufacturing where highly agile and accurate machines have been limited by their inability to adjust to changes in their tasks. Enabled to sense all six degrees of freedom, these machines will be able to adapt to new and complicated tasks without human intervention or delay.





Key Words: manufacturing, robotics, R&D 100 Award, six-degrees-of-freedom (SixDOF) sensor.

For further information contact Charles Vann (510) 423-8201 (vann1@llnl.gov).


Back to October 1996
Continue on to next R&D 100 winner