A developmental technique for automated alignment of threaded fasteners involves the use of the axial force between the fasteners as an indication of alignment. The technique was conceived as a means to guide a robot that is required, for example, to join a bolt with a nut.
The technique is based on the well-known fact that when a bolt and nut are properly aligned, gently pushed together along their common axis, and turned in the loosening direction (counterclockwise for conventional right-handed threads), there is a click - that is, a brief relaxation and recovery of axial force - once per rotation, each time the ends of the threads slip past each other. The technique is also based on the conjecture that the magnitude of the click decreases as the angle of misalignment between the bolt and nut increases.
In a typical application, the robot hand would hold the bolt or nut and would bring it into contact and approximate alignment with its mating nut or bolt, respectively. The robot hand would apply a small preload contact force. Then by actuation of the robot hand or by another mechanism cooperating with the robot, the bolt and nut would be rotated, relative to each other, in the loosening direction. During the rotation, strain gauges in the robot hand would measure contact forces that could be resolved into the axial contact force between the bolt and nut. The axial-force signal would be processed by the robot control system to determine the degree of bolt/nut misalignment (if any) and thus to determine any needed corrections to the position and orientation of the robot hand.
In experiments to test this concept, a robot hand held a 1/2-in. (12.7-mm)-diameter bolt against a nut that was rotated counterclockwise with a period of about 18 seconds. The left part of the figure shows an example of the axial force versus time in an experiment in which the angle between the bolt and nut axes was ≤1°. The initial rise in force of 1.3 lb ( ≈5.8 N) was caused by a command from the robot control system to press the bolt into the nut with this amount of force. (This level of preload was essential for reducing the effect of strain-gauge noise.) At approximately 5 seconds, the counterclockwise motion started. At approximately 21 seconds, there was a 0.5-lb (2.2-N) dip in axial force (a click), indicative of proper alignment of the nut and bolt threads. At this point, the robot control system sensed the change in the axial force and the rate of change of axial force, and responded by generating a command to reverse the rotation in order to tighten the bolt and nut. Thereafter, the axial force decreased as the engagement of the nut and bolt relieved some of the preload.
The right part of the figure depicts the axial force versus time in an experiment that was similar except that the angle between the bolt and nut axes was between 4 and 6° (close to the cross-threading angle for the particular bolt and nut). The initial rise in the axial force was similar to the one described above, but the subsequent clicks were smaller. The force and the rate of change of force did not change sufficiently for the robot control system to recognize "good" alignment; therefore, the system did not command a reversal of rotation from counterclockwise to clockwise. However, the axial-force signal included a periodic feature that indicated the instant when the threads could mate if they had been better aligned. In principle, the robot control system could be modified to recognize this feature and to command the fingers of the robot hand to tilt the bolt as a function of an axial-force gradient until alignment was obtained.
This work was done by Myron A. Diftler and Michael L. Ross of Lockheed Martin for Johnson Space Center. MSC-22837