Electricity: motive power systems – Positional servo systems – Program- or pattern-controlled systems
Reexamination Certificate
1999-07-19
2001-10-16
Nappi, Robert E. (Department: 2837)
Electricity: motive power systems
Positional servo systems
Program- or pattern-controlled systems
C901S003000, C901S010000
Reexamination Certificate
active
06304050
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a means and method of determining the internal robot joint sequence whose execution will cause any of a variety of precisely specified rigid-body actions of a robot-controlled tool relative to an arbitrarily positioned, oriented and shaped workpiece, without prior mathematical characterization of the workpiece geometry/position/orientation, based upon laser-pointer-assisted vision, without calibration either of the visual means or of the robotic means or of the laser-pointer means. The invention also includes a means by which a user, with a graphical user interface, may specify task objectives in a single computer-monitor image of the workpiece surface of interest, whereupon the robot executes the three-dimensional task autonomously, precisely and without calibration.
PROBLEMS IN THE ART
The present invention remedies the problem of “wasted” or “unusable” mechanical dexterity by providing a robust and precise means to control that dexterity relative to arbitrarily located workpiece surfaces. Robots with five, six or more degrees of freedom, now commonplace, typically have the mechanical dexterity needed to accomplish a very wide array of practical three-dimensional tasks, with extremely high precision, on arbitrarily located/oriented workpieces. Nevertheless, this dexterity is seldom applied to the broad range of mechanically possible ends because of the lack of an effective, practical control technology which the present disclosure discusses. “Dexterity”, in this context, refers to the intrinsic, mechanical ability to bring about the desired series of tool position/orientation poses, as required for the task at hand, by means of servo control of the individual, internal degrees of freedom (typically joint rotations) of the robot.
For example, consider the task of etching or routing a figure, using a rotating tool, into an arbitrarily located surface: If either the workpiece, robot, or both are moved such that the workpiece surface is adequately within the robot's workspace or “reach”, and if the appropriate etching tool is rigidly fixed to the robot's outermost (end) member, there will exist a sequence of easily trackable joint rotations, internal to the robot, which will complete the etching, with extremely high precision, as the task has been specified relative to the surface.
In practice, however, absent the presently disclosed art, it is prohibitively difficult and time-consuming to identify the requisite robot joint-rotation sequence; it is seldom “worth the trouble” to use a robot for a single workpiece/robot repeat cycle, or for small-lot operations, and therefore the aforementioned dexterity is not exploited. The same intrinsic mechanical ability to utilize robots, but control impracticality, pertains to a wide variety of surface-relative tasks which require dexterity, including cutting, drilling, all kinds of surface preparation (cleaning, sanding, scouring, painting), grasping of objects of known geometry, certain kinds of assembly, digging, excavating, fastening, gluing, laying down adhesives, welding, inspection, surgical applications and so on.
Where the previously mentioned task, etching or routing, is automated with existing technology, it is overwhelmingly commonly the case that the workpiece is constrained or fixtured such that the surface of interest has a precisely established position and orientation relative to a low-degree-of-freedom, planar, highly calibrated control system which guides the etching or routing tool. This need for a fixturing operation precludes applications where the workpiece cannot be accommodated within the fixture for any of a variety of reasons including workpiece size, shape, or location.
Because of the lack of an effective, practical robot-control technology—which technology the present work discloses—effective automation of the etching task is often precluded. As mentioned, however, insofar as mechanical dexterity is concerned, many common robots in existence today are mechanically capable of the requisite motion if placed adequately near to the workpiece surface; the limiting aspect of sensing and control extends beyond etching and routing to the variety of other tasks mentioned above; therefore, the prospects for expanded use of dexterous robots via the present enabling technology is enormous.
Most present, practical, 3-dimensional applications of robots make use of the “teach-repeat” mode of operation; here, the requisite joint rotations for a given task/workpiece location are first established (“taught”), usually by a human operator, and these sequences are “blindly” repeated with each consecutive workpiece by using servo control at each joint in order to recover the taught internal joint configurations. This not only precludes moving the robot base into the environment of a fixed workpiece (e.g. a building facade or an aircraft); it also requires that each new workpiece be precisely repositioned as the prototype. The present art allows for movement between operations of the robot base, and does not require the precise location of the workpiece within the robot's workspace preceding each repeat cycle.
Types and Deficiencies of Existing Methods of Robot Supervision and Control (other than the “teach-repeat” mode)
One example of particular need for robot application is space operations. Where robots have been used to engage satellites or manipulate objects in orbit, “teleoperation” by astronauts who are also in orbit and who generally have direct visual access both to the robot and the object of interest, has been applied.
Teleoperation has many proposed forms, but its meaning is restricted in this discussion to those systems where actuation of all degrees of freedom of the robot are directly controlled in real time by the human teleoperator. The exact means of human input can vary from kinematically similar exoskeletons, wherewith the robot mimics the motion of the operator, to joystick input where there is a relationship between rate and direction of the motion of the robot and various deflections of the joystick.
The practical problem encountered with this approach is related to the need for clear visual access to the pertinent manipulation event. Significant loss of capability is associated with all forms of indirect (e.g. video) transmission of visual information to the operator; this problem is worsened dramatically with even slight delays in signal transmission between the remote system and the teleoperator. Where such delays are on the order of minutes, such as, for example, with proposed Earth-to-Mars teleoperation, meaningful remote control is virtually impossible with teleoperation.
Related to this problem is the fact of the two-dimensional nature of any image. It is very difficult for an operator to make accurate motion judgments regarding the three-dimensional world with feedback from a two-dimensional image. This problem remains even if the user is presented with a stereo rendering of the scene acquired from a pair of parallel cameras; and it tends to be of relatively little help when the user is presented with two images from different camera perspectives.
Another proposed form of human supervisory control of remote robots, “virtual reality”, suffers from a kind of problem that is also more generally part of the central, very practical difficulty which has been encountered in extending the use of robots beyond the simple “teach-repeat” mode of operation—the problem of “calibration”. With the virtual-reality-based systems which have been proposed, the user is presented with visual responses to his motions which are geometrically identical to those which the actual physical system would present. The theory goes, then, that, by creating maneuver success in the virtual world, the operator creates simultaneous success in the remote, physical world.
Success with this approach is predicated upon (virtually) perfect calibration of the remote system, such that the underlying, modeled geometric relationships which permit correspondence between the actu
Gonzalez Galvan Emilio J.
Robinson Matthew L.
Seelinger Michael J.
Skaar Steven B.
Duda Rina I.
Nappi Robert E.
Neyerlin Wallace F.
Skaar Steven B.
LandOfFree
Means and method of robot control relative to an arbitrary... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Means and method of robot control relative to an arbitrary..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Means and method of robot control relative to an arbitrary... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2575372