Robotic arm

Description Improved Robotic

Technical Field

The present invention relates'generally to robots 5 and robot control and, more particularly, to partial open-loop control o£ tendon manipulated robot arms.

Background Art

In the past, robotic arms have been constructed and designed for use as either expensive, complex and 10 heavy-duty arms for heavy industrial use, or inexpensive, light-duty arms for educational and robotic research purposes. With respect to the industrial arms, closed-loop control systems are typically utilized with drive motors and shaft encoders often located 15 at each of the joints. Typical of these robots are those manufactured by Unimation Corporation and Cincinnati Milacron Corporation. The closed-loop configuration, as well as the use of shaft encoders and drives at each joint, permits an arm in which precise 20 control can be realized. However, due to the added weight at each joint, the structural components must be strengthened accordingly. In turn, the drive mechanisms must al$o be increased in capacity. As a result, such heavy-duty robots are expensive, bulky 25 and of great size. Conversely, with respect to the light-duty, educational type robotic arms, such as those arms manufacturered by the assignee of the present invention, the drive motors for each segment of the arm 30 are located at the base of the arm, with the motive energy being transmitted to each joint by way of a cable and pulley system, utilizing tendon technology. This, in turn, permits each segment of the arm to be of ® - (H significantly lighter weight. In order, to. keep costs down, such an arm is controlled open-loop. That is, the position of each member is controlled by keeping track of the signal supplied to the motor drives, with the assumption that the particular member will move exactly in accordance with the drive signal supplied thereto. Such a configuration provides an inherent problem, that of keeping track of the actual position, versus the commanded position, of the arm, and each of its structural members. With open-loop control, the arm is susceptible to mispositioning due to a number of factors, such as an external force which causes the arm to be bumped out of its current position and into another different position. The control circuitry for this open-loop configuration will have no way of detecting that such an event occurred. Additionally, as is the case in most mechanical motion involving movement over a series of different paths, the cumulative error in and of itself may result in a positional difference from that which has been commanded. Another source of mispositioning is slippage of the motor drives. For example, when stepper motors are used, such stepper motors are susceptible to slippage when a force greater than the maximum torque suppliable by the stepper motor is presented by an external object.

With respect to the tendon-type robotic arms in which drive motors are mounted on the base of the arm, there is an additional problem in designing a position 30 calibration system. This problem involves the intercoupling between each of the segments of the arm.

The typical robotic arm is comprised of a number of structural members, each structural member being linked to another by way of a joint. The cables or tendons by 35 which rotational energy is transferred, from the motor . . ¢1 s drives on the base to the structural members, are themselves routed to the desired structural member via other structural members and joints. As such, the movement of a particular structural member can affect the position of other structural members. This, in part, is due to the manner in which the cables pass through a particular structural member on their way to the structural member to be driven. Typically, an idler pulley is located on an intermediate joint. The cable is wrapped about this idler pulley. The structural member driven by that cable is positioned at the end of the intermediate structural member which pivots about the joint on which the idler pulley is positioned. Thus, when that intermediate structural member is moved, and where the motor drive for the idler pulley does not move, the movement of the intermediate structural member causes the cable to wrap or unwrap an additional amount about the idler pulley. To the structural member to which the cable is connected, this wrapping or unwrapping looks like the motor drive is causing the cables or tendons to be moved. Thus, this structural member moves when the intermediate structural moves. Conversely, certain of the structural members do not influence the position of other structural members. Typically, these members are located at the end of a series of coupled members. This is because there are no idler pulleys associated with other structural members on the joint to which these end members are connected.

As a result of the intercoupling of the various structural members, the manner in which the position of each structural member is calibrated is no longer straight-forward. Any calibration system design must take into account this intercoupling and deal with it effectively.

<img file="IL69862A_D0001.tif" />

Summary of the Invention

The foregoing and other problems of prior art robotic arms are overcome by the present invention of a method and apparatus for manipulating a tendon con5 trolled robotic arm in a multi-axis, open-loop system of coupled structural members, wherein each structural member is pivottably coupled to another of the structural members to form a number of joints, for positioncalibrating each structural member comprising position sensor means located at each joint for providing a home signal when the structural members associated with the joint are in a predetermined orientation; and control means responsive to the home signal from each position sensor means for modifying the position of each struc15 tural member in a predetermined order and by predetermined positional increments so that each of the structural member is caused to be positioned with respect to one another such that all position sensors simultaneously provide home signals.

In one embodiment of the present invention, each of the structural members are controlled by one or more separate motors. The motor is coupled to the particular structural member by way of a cable/pulley drive train. The present invention provides position calibration of the various structural members with respect to one another by providing drive signals to each of the motor/cable/pulley drive arrangements in response to the output states of the position sensors. The first step of the procedure is to simultaneously modify the position of all structural members in the direction of the home position of each structural member until a change in state for each of the associated position sensors is encountered. The next step is to move the structural members in an ordered sequence, one by one, until each structural member has encountered its home position. The final step is to move each structural member, in an ordered sequence, one predetermined position increment away from its home position, to verify that the home position has been encountered.

The present invention also provides an apparatus 5 and method for securing and routing the various drive cables to the pulley structures so that a more positive coupling between the motors and the structural members can be obtained.

Also provided is a position sensor structure by 10 which the position of a cable with respect to a structural member can be sensed.

It is therefore an object of the present invention to provide an apparatus for position calibrating a multi-axis, open-loop system of coupled structural 15 members.

It is a further object of the present invention to provide an apparatus for position calibration of a multi-axis, open-loop system of coupled structural members, wherein each structural member is driven by a 20 motor/cable/pulley arrangement.

It is another object of the present invention to provide an apparatus for position calibration of a multi-axis, open-loop system of coupled structural members, wherein each structural member is driven by 25 a motor/cable/pulley arrangement and futtherwherein each of the pulleys are grooved in a sequence of alternating right and left handed pitches in order to guide the cables thereon, and furtherwherein the cables are positively anchored to each pulley by way of a fitting which has been swaged to the cable and inserted into a hole in a slot in each pulley.

The foregoing and other objectives, features and advantages of the invention will be more readily understood upon consideration of the following detailed description of certain preferred embodiments of the invention, taken in conjunction with the accompanying drawings, a

Brief Description of the Drawings

Figure 1 is a simplified block diagram of the robotic arm and control system of the present invention.

Figure 2 is a simplified depiction of the robotic arm of the present invention, illustrating the motor/ cable/pulley drive arrangement.

Figure 3 is a rear view of the robotic arm of the present invention, illustrating the routing and positioning of the various pulleys and cables on two of the joints.

Figure 4 is a diagram illustrating the arrangement 15 of the pitch of the grooves for each pulley in a sequence of pulleys according to the present invention.

Figure 6a illustrates the positioning of a spherical member upon a cable in accordance with the present invention*

Figure 6b is a cross-section of the spherical member of Figure 6a taken along lines 6b-6b.

Figure 6c illustrates the positioning of the swaged spherical member and cable with respect to a pulley.

Figure 6d is a top view of the spherical member which has been swaged to a cable and positioned in a slot on a pulley in accordance with the present invention.

Figure 5 illustrates the cabling and pulley arrangements for the various structural members of the present invention.

Figure 7a illustrates one embodiment of a position sensor.

<img file="IL69862A_D0002.tif" />

-רFigure 7b illustrates a cross-section of the position sensor or Figure 7a, taken along lines 7b7־־b, Figure 8 is a cut-away view of an embodiment of a position sensor for sensing the position of a cable.

Figure 9 is a simplified block diagram of the method of position calibrating of the present invention.

Figure 10 is a more detailed block diagram of the method of position calibration of the present invention.

Detailed Description of the Invention

Referring to Figure 1, the basic components by which the present invention is implemented will now be briefly explained.

The robotic arm 10 is controlled by an intelligent controller 12. The intelligent controller 12 provides 15 motor drive signals to motors 14, such as stepper motors shown mounted on the body 16 of the arm 10, via motor driver circuitry 22.

The robot arm 10 is comprised of several structural members. A base 24 is provided to support the 20 other structural members. A body 16 is mounted for rotation in a plane horizontal to the plane of the base by way of base joint 26. Such a joint can be considered bo be a swivel .joint. Upper arm 28 is mounted at one end to body 16 by way of a shoulder 25 joint 30. This shoulder joint can be termed a hinge joint. The other end of upper arm 28 is coupled to forearm 32 by way of elbow joint 34. Elbow joint 34 can also be described as a hinge-type joint. Hand or gripper 36 is coupled to forearm 32 by way of wrist 30 joint 38. Wrist joint 38 is a differential joint.

The differential joint which couples hand 36 to forearm 32 permits hand 36 to be positioned with respect to forearm 32 in any position within a partial sphere which extends from at the top of forearm 32, in an upward arc therefrom, and terminates toward the bottom of forearm 32. Hand 36 can be rotated anywhere within this partial sphere.

The elbow joint 34 permits forearm 32 to be pivotted in a common plane with respect to upper arm 28 over approximately 180 degrees.

Shoulder joint 30 permits upper arm 28 to be . pivotted in a common plane, with respect to body 16, over approximately 180 degrees. Finally, base joint.16 permits body 16 to be rotated with respect to base 24 in a plane horizontal to base 24 over nearly 360 degrees.

Referring to Figure 2, the manner in which motors 14 control the position of each of the structural members will now be described. For purposes of explanation, a single motor 14 is shown. Also, assume that motor 14 is a stepper motor. It is to be understood that, in practice, there will be a plurality of stepper motors with each stepper motor being assigned to control a particular structural member.

Stepper motor 14 is mounted to body 16 and coupled to a timing .belt gear 40 via a timing belt 42. Timing belt gear 40 is mounted for rotation on a common drive shaft 44. The timing belt;gears associated with the other stepper motors are also mounted on this common drive shaft 44. a drive capstan 46 is coupled for rotation with timing belt gear 40 and also mounted bn common drive shaft 44. See Figure 3. Thus, as stepper motor 14 rotates, timing belt 42 transfers the rotational motion to timing belt gear 40. As timing belt gear 40 rotates, capstan 44 is caused to rotate about common drive shaft 44. Assume for purposes of explanation that the particular stepper motor 14 shown is assigned to drive the cables for movement of forearm 32. A cable 48 is wrapped a predetermined number of

<img file="IL69862A_D0003.tif" />

times about capstan 46. One end of cable 48 is then wrapped about idler pulley 50 from the top. Idler pulley 50, is supported for rotation by shoulder axle 52, which in turn forms a part of shoulder joint 30.

The other end of cable 48 is wrapped about idler pulley 50 from the bottom thereof. The ends of cable 48 are then extended through upper arm 28 and wrapped about drive pulley 54. Drive pulley 54 is rotatably mounted on elbow axle 56 and fixedly secured to the end of forearm 32. As before, one end of cable 48 is wrapped about drive pulley 54 from the top, while the other end of cable 48 is wrapped about drive pulley 54 from the bottom. The free ends of cable 48 are then secured to forearm 32. Elbow axle 56 forms a part of־ 15 elbow joint 34.

Thus, as capstan 46 is caused to rotate by the rotation of timing belt gear 40, such rotation is transferred to the elbow drive pulley 54 via cable 48 and idler pulley 50. Because drive pulley 54 is 20 fixedly attached to the end of forearm 32, the rotation of drive pulley 54 causes forearm 32 to pivot about elbow joint 34.

In a similar manner, a stepper motor 14, timing belt, timing gear, capstan־, idler pulley, drive pulley, 25 and drive cable configuration is utilized to drive each of the other structural members. From Figure 3, the positioning of the timing belt gears on common drive shaft 44 can be seen. Also, the positioning of the various idler pulleys and drive pulleys on shoulder axle 30 52 can also be seen. Finally, Figure 3 illustrates the positioning of various idler pulleys and drive pulleys on elbow axle 56. Also shown is the routing of the various drive cables between the capstans, idler pulleys and drive pulleys between the common drive 35 shaft 44, shoulder axle 52 and elbow axle 56.

From Figure 3, it can be seen that timing belt gear 58 causes capstan 60 to rotate. This rotational motion is transferred via drive cable 62, to shoulder drive pulley 64. Shoulder drive pulley 64 is fixedly attached to upper arm 28 and rotatably mounted on shoulder axle 52.

Similarly, from Figure 3, it can be seen that timing belt gear 66 drives capstan 68, which in turn controls drive cable 70. Drive cable 70 can be seen to wrap around idler pulley 72, which is rotatably mounted on shaft 52 of shoulder joint 30. Drive cable 70 extends through upper arm 28 to wrap around idler pulley 74, which is rotatably mounted on elbow axle 56 of elbow joint 34. From Figure 2, it can be seen that drive cable 70 extends from idler pulley 74 through forearm 32 to wrap around differential drive pulley 76. Differential drive pulley 76 can be seen more clearly in Figure 1. Rotation of differential drive pulley 76 in conjunction with the complementary differential drive pulley 78 acts to change the position of hand 36. Differential pulleys 76 and 78 are rotatably mounted on wrist axle 80 which is located in wrist joint 38. Differential pulleys 76 and 78 each drive a bevel gear. In turn, each bevel gear meshes with a bevel gear on the hand 36. By rotating differential pulleys 76 and 78 in the same direction, pitch is achieved. By driving differential pulleys 76 and 78 in the opposite direction, roll is achieved.

The cable which drives differential pulley 78 is shown in Figure 3, and is labelled with reference designation 82. This cable is driven by capstan 84 arid timing belt pulley 86. Drive cable 82 passes around idler pulley 88 which is mounted on shoulder axle 52 and around idler pulley 90 which is mounted on elbow axle 56.

Β Ο

Remaining with Figure 3, it can be seen that a stepper motor 14 drives a shaft 92 which, in turn, drives timing belt 94. Timing belt 94, in turn, drives timing belt gear 96, and hence capstan 98. Capstan 98, 5 in turn, drives cable TOO. Cable 100 is wrapped around idler pulley 102 which is rotatably mounted' on shoulder axle 52. Additionally, cable 100 is wrapped around idler pulley 104 which is rotatably mounted on elbow axle 56. Although not shown in Figure 2, drive cable 10 100 extends as a single length through forearm 32 to hand 36. Cable 100 is utilized to open and close hand 36.

Finally, from Figures 2 and 3, it can be seen that timing belt 106 is driven by a stepper motor to drive 15 capstan 108. In turn, capstan 108 drives cable 110.

Cable 110 is routed to base pulley 112 via idler pulley 114. The purpose of idler pulleys 114 is to change the direction of cable 110 so that the cable feed is tangent to the surface of both capstan 108 and base 20 drive pulley 112.

By utilizing a timing belt to transfer the rotation of the stepper motor 14 to the timing belt gear, the amount of gearing backlash in the gearing is significantly reduced.

Structuring of the drive mechanism as described above results in a multi-axis system of coupled structural members. What is meant by this terminology is that the movement of certain of the structural members will cause a resulting movement of other structural 30 members. Conversely, the movement of certain ,different structural members will not influence the position of other structural members.

In order to better understand this relationship, reference is made to Figure 2. in the system of 35 structural members as interconnected as described above, movement of the structural member represented by the upper arm 28 about shoulder joint 30 will result in a change in orientation of forearm 32 with respect to upper arm 28 and a change in orientation of hand 36 5 with respect to forearm 32. Similarly, a change in position of forearm 32 about elbow joint 34 will result in a change in orientation of hand 3 with respect to forearm 32, ,as well as a change in the opened/closed condition of hand 36.

Conversely, a change in position of hand 36 with respect to forearm 32 will not result in any change in position in any of the other structural members. Thus, it can be seen that there is, what can be loosely termed, a unidirectional coupling of the structural members. With respect to the drive configuration described above, this coupling is considered to be in an outward direction, where the direction is referenced to the base 24.

The reason for such coupling can be best explained 20 by way of example. Referring to cable 48, associated with drive pulley 54 and forearm portion 32, it can be seen that when the upper arm 28 is rotated about shoulder joint 30, and when capstan 46 associated with drive cable 48 remains stationary, the effect is to 25 wrap one of the segments of cable 48 about idler pulley 50. The other segment of cable 48 is unwrapped from around idler pulley 50 to a corresponding degree. From the point of view of elbow drive pulley 54, this wrapping and unwrapping is equivalent of a rotation by 30 capstan 46. Thus, elbow drive pulley 54 rotates in response to the movement of upper arm 28, thereby causing forearm 32 to move. In general, it can be said that where a drive cable is routed about an idler pulley located on a particular joint, and where the 35 orientation of the structural members associated with that joint are changed, the drive cable associated with the various idler pulleys on that joint will be wrapped around and unwrapped from those idler pulleys. This, in turn, will cause the structural members with which the drive cables are associated to move. Thus, because the shoulder axle 52 has a large number of idler pulleys mounted thereon, it follows that upper arm 28 is coupled to the greatest extent to the other structural members. From Figure 3, it can be seen that elbow axle 56 supports the idler pulleys for the differential drive of hand 36, as well as the open and close drive for the hand 36. Thus, movement of forearm 32 about elbow joint 34 will cause a correspending movement of hand 36. Conversely, since no idler pulleys are located on either base joint 26 or wrist joint 38, there is no coupling of the base or hand to the remainder of the structural members.

The above-discussed inter-relationship or coupling between the various structural members must be taken into account when deciding upon the position commands to be supplied to each stepper motor in order to obtain a predetermined orientation of the various structural members. The above relationship is also significant when position calibration of the various structural -*־ members is required. As such, position calibration is neither straight forward nor simple.

In implementing the above* described cable drive system, it has been found that certain refinements enhance the performance of the system. Among these refinements are the use of grooves in the drive pulley, the idler pulleys and the capstans, as well as a specific orientation of the grooves therein in relation to the grooves on the previous and subsequent pulleys or capstans in the drive sequence. More specifically, it has been found that, by providing a pulley which is

.''׳ס

W spirally grooved, cable life is increased. The grooving also helps to ensure that the cables remain on the pulley under normal operating conditions. In the past, side plates were required to be mounted with each pulley to prevent the cables from'falling off from the pulley during normal operation. With the use of a grooved pulley, this problem has been solved. Of course, where extraordinary load conditions are encounteredj such as collisions with external objects, flanges are useful in keeping the cables in place. It has additionally been discovered that in a capstan, idler pulley, drive pulley sequence, by alternating the pitch of the grooves so that the pitch changes alternatively in orientation from right to lefthanded from element to element within the sequence, cable wear can be minimized, and the grooves themselves made more shallow. For example, referring to Figure 3, and more specifically to capstan 46, idler pulley 50 and drive pulley 54. The grooves on each of these elements, when arranged in accordance with the present invention, vary from a righthanded pitch on the capstan 46, to a lefthanded pitch on the idler pulley 50, and to a righthanded pitch on the drive pulley 54.

This relationship is shown in greater detail in Figure

4. The advantage of arranging the pitch of the grooves in an alternating manner from pulley to pulley is that the cable can now exit and enter the grooves on the pulleys in better alignment therewith. As the cable exits from capstan 46, it extends in a substan30 tially straight line to a corresponding groove on idler pulley 54. Cable 48 is then wrapped around idler pulley 54 in a direction which is determined by the pitch of the grooves thereon. For the portion of cal]be 48 which enters idler pulley 54 from the top, the pitch of idler pulley 54, which is a lefthanded pitch, causes that portion of cable 48 to be wrapped toward the interior of idler pulley 54. Similarly, for the portion of cable 48 which enters from the bottom of idler pulley 54, the cable is also wrapped toward the interior of idler pulley 54.

If, on the other hand, idler pulley 54 were of the same pitch orientation as capstan 46, the cable portion which enters idler pulley 54 from the top would be wrapped in a direction toward the exterior of the pulley. Similarly, the portion of cable 48 which enters idler pulley 54 from the bottom would be wound in a direction toward the exterior of the pulley. It can be seen that, in order to accommodate both sections of cable 48 on an idler pulley having the same pitch orientation as the pulley or capstan feeding it, both cable portions must enter the idler pulley at its center so that they can then be wound, according to the pitch, toward the exterior of idler pulley 54. With this requirement it can be seen that the angle of entry is much more accute than when an alternating pitch direction is used. The more accute entry angles of the non-alternating pitch arrangement results in greater cable wear due to the rubbing of the cables against the sidewalls of the grooves. Additionally, the grooves themselves must be־made deeper in order to maintain the 25 cables within the grooves.

On the other hand, because an alternating pitch arrangement permits the cable to enter and exit the grooves of a pulley in better alignment with the grooves, cable wear is significantly reduced, and 30 the depths of the grooves can also be reduced.

In Figure 5, the location and groove pitch of each of the pulleys is shown. Capstan 60 and shoulder drive pulley 64 both have a righthanded pitch. This is because only one end of cable 62 is wrapped around 35 shoulder drive pulley 64. The other end of cable 62 is positioned over a small portion of shoulder drive pulley 64. As can be seen from Figure ’5, the ends of drive cable 62 are secured to the upper arm 28.

With respect to the drive arrangement for differential drive pulley 76, capstan*68 has. a lefthanded 5 pitch, idler pulley 72 has a righthanded pitch, idler pulley 74 has a lefthanded pitch, and differential drive pulley 76 has a righthanded pitch. As such, drive cable 70 from capstan 68 enters idler pulley 72 from the exterior thereof and exits therefrom from the interior of drive pulley 72. Cable 70 from idler pulley 72 enters idler pulley 74 from the interior thereof and exits from the exterior thereof. Finally, cable 70 from idler pulley 74 enters differential drive pulley 76 from the exterior of the drive pulley. It is 15 to be noted that the ends of drive cable 70 are joined together at a joint 116. The ends of cable 70 are joined in this manner for purposes of convenience, there being no convenient point on drive pulley 76 to which the cable ends can be secured.

The drive linkage for differential drive pulley 78 is similar to that for differential drive pulley 76, except that the pitch variation is from a righthanded pitch at capstan 84 to a lefthanded pitch at idler pulley 88 to a righthanded pitch at idler pulley 90 and 25 finally to a lefthanded pitch at differential drive pulley 78.

\ Referring to the middle of Figure 5, cable 100, which opens and closes hand is shown as a single length which is wrapped around capstan 98, idler pulley 30 102, and idler pulley 104. The grooves for all of these elements are orientated in a righthand pitch. The other end of cable 100 is shown anchored to the drive gear 96, By providing a groove pitch of the same orientation for this drive link, the position of the 35 cable can be shifted toward the center of the robotic arm as the cable is extended therethrough. Note that timing belt gear 96 and capstan 98 are positioned off-center and on common drive shaft 44. Thus, there is a need to shift the position of* cable 100 towards 5 the center of the arm 10 so that cable 100 will be centered when it enters hand 36. The righthand pitch of the grooves on idler pulley 102 and idler pulley 104 cause the cable, as it is wrapped around each idler pulley, to progress from the left to right and thereby 10 towards the center of the arm.

Finally, as was shown in simplified form in Figure 4, it can be seen from Figure 5 that the drive linkage for elbow drive pulley 54 provides for a righthand capstan 46, a lefthand idler pulley 50, and a righthand 15 drive pulley 54.

Figures 6a through 6d illustrate a further feature of the present invention. It has been discovered that, by swaging a spherical ball onto each drive cable and then positioning this swaged ball in a slot in the 20 capstan associated with the drive cable, that a more positive drive capability and easier assembly is achieved. This is because positioning of the ball within the groove fixes the position of the cable with, respect to the capstan. This procedure is also util25 ized for positively positioning a cable with respect to each of the idler pulleys.

Figure 6a shows the positioning of the ball 118 on a drive cable. Figure 6b illustrates a cross-section of ball 118 taken along lines 6b-6b of Figure 6a.

Figure 6c illustrates the positioning of the swaged ball within slot 120 which has been cut into a capstan or idler pulley. Figure 6d is a top view of the swaged ball as it is positioned within a slot 120.

In practice, each of the cable portions which are 35 wrapped around a capstan will be.anchored to the capstan using a swaged ball arrangement. Similarly, ® c וwhenever a cable is wrapped around an idler pulley, a swaged ball 118 and slot 120 arrangement will be used. Finally, with respect to the differential drive pulleys 76 and 78, such a swaged arrangement is also 5 be used. This is because the ends of the cables associated therewith are not anchored to a structural member, but rather tied to one another.

As discussed previously, control of the system of the present invention is generally open-loop. That is, 10 drive signals are supplied to motors 14 which turn the various capstans for movement of the various drive cables. As result of the drive cable movement, the desired structural members are changed in position. It is asssumed that each structural member will move in 15 accordance with the drive signals supplied to its associated motor 14. Thus, errors can result in the position of the structural members when the motor 14 slips or when an external force causes the particular structural member to be moved out of position. Because 20 no shaft encoders are utilized, there is no straightforward manner in which the actual position of each structural member can be easily obtained. The determination of structural member position is further complicated by the coupling between the various struc25 tural members as discussed above.

In the present invention, there is provided a method and apparatus for calibrating the position of each structural member. There is provided in association with the structural members which rotate about 30 each of the Various joints, a position sensor which . provides an output which has one of two states. The transition between these two states is treated as a reference or home signal for the associated structural members for a particular joint. By positioning 35 each of the structural members so that the position sensors.associated with each joint all indicate a home signal, a reference or home position can be obtained. All subsequent movement of the various structural members can then be referenced to this home position. Additionally, any mispositioning of the various structural members can be determined by'counting the number of commands or steps actually required to position a structural member into its home position and by comparing that number with the theoretical number of commands which should have been required to place the particular structural member into its home position.

Thus, upon the initial power-up of the robotic arm, or periodically during the operation of the arm, a homing mode can be entered into by which the arm can calibrate its position, determine any position errors, and thereby correct any mispositioning.

Referring to Figures 7a and 7b, a typical position sensor will now be described. Figure 7a is a simplified view of a position sensor which can be disposed on the elbow joint 34. A .portion of forearm 32 is shown mounted on elbow axle 56 with respect to a portion of upper arm 28. A tab 122 is shown positioned on forearm 32 at a predetermined radial distance from elbow axle 56. A light emitting diode (LED) phototransistor element-124 is shown positioned on upper arm 28 at the same predetermined radial distance from elbow axle 56. Figure 7b is a view of the above-described arrangement taken along line 7b-7b of Figure 7a. LED/phototransistor element 124 comprises a LED portion 126 and a phototransistor portion 128. These portions are aligned with one another and spaced apart by a gap 130. When gap 130 is clear, a light beam from LED portion 126 spans gap 130 and is detected by phototransistor portion 128. When forearm 32 and upper arm 28 are rotated with respect to one another, such that tab 122 is caused to enter gap 130, the light beam from LED 126 is blocked. The output of LED/phototransistor element 124 will be in one state when gap.130 is clear, and will be in another different state when tab 122 enters gap 130. The transition between the two different states is treated as an edge or state transition and plays a key role in the method of position sensing of the present invention.

Similar position sensors are disposed on the shoulder joint 30 and the base joint 26. It is to be understood that other forms of position sensors can be utilized with satisfactory results. Such sensors include microswitches, and the like.

Due to the arrangement of the cables which drive the differential drive pulleys 76 and 78, a different position sensor is provided for the wrist joint. Figure 8 illustrates such a sensor.

As discussed above, the ends of drive cable 70 are tied to one another at junction 116. See Figure 5. Similarly, the ends of drive cable 82 are tied to one another at junction 117. Typically, these junctions take the form of turnbuckles which permit easy adjustment of the tension of the cables. In Figure 8, turnbuckle 132 can be seen to join the ends of cable 70 together, while turnbuckle 134 can be seen to join the ends of cable 82 together.:: Recall from the above discussion that cable 70 is fastened to differential drive pulley 76 by a swaged ball disposed in a slot within differential drive pulley 76, and that cable 82 is fastened to differential drive pulley 78 in a similar manner. Thus, the position of each of the differential drive pulleys 76 and 78 can be related directly to the position of cables 70 and 82.

Thus, it follows that a sensing of the position of drive cable 70 and drive cable 82 will provide an indication of the position of each of the differential drive pulleys 76 and 78.

® W

In order to provide such a position sensing of the cable positions, the arrangement of Figure 8 is provided. Figure 8 shows a cutaway portion of forearm 32. Cables 70 and 82 are shown parsing through this portion. These cables 70 and 82 both pass through apertures within plates 136 and 138. Plate 136 is pivottably mounted to forearm 32 by way of shaft 140, while plate 138 is pivottably mounted to forearm 32 by shaft 142. Shaft 144 provides a pivot-limit to plate 136 while shaft 148 provides the pivot-limit to plate 138. A spring 150 is connected between the tops of plates 136 and 138 to bias the tops of the plates towards one another. The plates 136 and 138 are located with respect to one another and with respect ־ of drive cables 70 and 82 so that joints 116 and 117 are located between the plates 136 and 138. Positioned at joint 116 is a disc 152, while positioned at joint 117 is a disc 154. Each of these discs is positioned at right anglesto their associated cables.

Plates 136 and 138 also serve as mechanical stops for each of the cables; thus, plate 138 serves as a mechanical stop for cable 70, while plate 136 serves as a mechanical stop for cable 82. In the position sensor illustrated in Figure 8, the position of each cable associated with the end of the cable travel is chosen as the position to be sensed. Thus, plate 138 is shown positioned with respect to disc 152 on joint 160 so that when disc 152 engages with plate 138, cable 70 is at the end of its desired travel. Likewise, plate 136 is shown positioned with respect to joint 170 and disc 134 such that when disc 134 engages with plate 136, cable 82 is at the end of its desired travel.

Shown positioned on plate 138 is a tab 156 and a LED/phototransistor element 158, which function in a manner similar to the position sensor described in connection with Figures 7a and 7b. when disc 152 is disengaged from plate 138, spring 150 biases plate

138 such that tab 156 is separated from LED/phototransistor 158. Conversely, when disc 152 engages plate 138, tab 156 interrupts the light beam within the gap of LED/phototransistor 158. A similar tab and LED/phototransistor arrangement is disposed toward the bottom portion of plate 136. In order to simplify the figure, the tab and LED/phototransistor structure associated with plate 136 is not shown. In operation, the tab of this structure is disposed within the gap of the LED/phototransistor element when disc 154 is disengaged from plate 136. Conversely, when disc 154 is engaged with plate 136, the tab is removed from the gap of the LED/phototransistor element. Thus, the15 signals provided by each of these structures are opposite in polarity.

Preferrably, the position sensors are located on the structural members such that only one state transit tion can occur in the range of motions for the particu20 lar structural members. Thus, in this configuration, the state of the position sensors also indicates in what direction the structural members should be manipulated in order to produce or attain the state transition or home position. . :;

The manner in which the outputs from these various position sensors is utilized to provide position calibration will now be explained in greater detail. Referring to Figure 9, a method of utilizing the signals from the above-described position sensors to calibrate the position of the robotic arm will now be described. In general, the method involves three stages, the first of which is shown in step 160. In this step, drive signals are provided, in parallel, to all of the motors 14 until a transition in the output state for each position sensor is detected. These transitions are also called edges. When a particular sensor outputs, a transition, the associated structural member is caused to move back and forth across this transition. When all of the position sensors have provided an output state transition, step 162 is 5 executed.

In step 162, the motors 14 are driven in an ordered sequence. This ordered sequence begins with the driving of the motor 14 associated with the structural member having the highest degree of coupling to the 10 other structural members. Recall from the discussion above, because of the use of idler pulleys and their relationship to the various joints of the robotic arm, the movement of certain of the structural member about a particular joint, will cause movement of other structural members about a different joint. Thus, in the configuration of the present invention, wherein an outward coupling is provided, the hand 36 would have the least amount of coupling. Additionally, because there are no idler pulleys associated with the base joint 26, other than the idler pulleys for the base drive cables 110, the base 24 itself has a low degree of coupling to the remainder of the structural members. The forearm 32 has the next lowest degree of coupling with respect^:to the other structural members. Finally, 4;

the upper arm 28 has the highest degree of coupling to the other structural members. Thus, step 162 would operate upon upper arm 28 initially, followed by forearm 32, and then hand 36. It is to be understood that this sequence will change if the coupling between the structural members changes. Thus, if there were inward coupling, such that the hand 36 had the highest degree of coupling and the upper arm 28 had the lowest degree of coupling, the hand 36 would be processed first and the upper arm 28 would be processed last.

In step 162, then, each motor 14 would be driven in the ordered sequence, until its associated position sensor indicated that a home state had been reached. As used herein, home state is defined as an arbitrary 5 state of a particular position sensor, the state being determined by the output state of the position sensor . on one side of the edge or transition. Recall that each position sensor supplies one of two output states, and that the transition between these states is of key 10 interest. Because it is difficult to maintain the position of a structural member so that the associated position sensor is at its output transition, a position to one side or the other of the transition is designated as the home state. Thus the home state for 15 a particular position sensor corresponds to the sensor output state for a position a small distance removed from the transition position, and in a predetermined direction. For example, with respect to the position sensor shown in Figure 7a and Figure 7b, the home state 20 would be chosen as that which indicates that the tab 122 is positioned within gap 130. Conversely, with respect to the position sensor associated with plate 136 of Figure 8, the home state would be chosen to correspond to the output of the LED/phototransistor 25 element when the tab is removed from gap.

After each of the motors 14 has been driven to cause the associated position sensors to assume their home state, step 164 is executed. In step 164, the motors 14 are driven, in the same ordered sequence as 30 in step 162, to cause their associated structural members to move in the opposite direction by a predetermined amount until the position sensor. makes a transition out of its home state. The purpose of step 164 is to ensure that each of the joints is positioned 35 within a preselected incremental position of the !f-Λ , ' transition point of the position sensors. Thus, after step 164 has been successfully completed, all of the structural members can be said to be within one incremental unit of their home position.

Referring to Figure 10, the h'oming method will be described in greater detail and in conjunction with the intelligent controller 12 and motor drive circuitry shown in Figure 1. As discussed previously, the structural members of the robotic arm 10 are controlled 10 by a cable and pulley system which are, in turn, driven by motors 14. Motors 14 receive drive signals from motor driver circuitry 22. In turn, motor driver circuit 22 receives instructions from an intelligent controller 12. This intelligent controller 12 can be15 a general purpose microcomputer or a dedicated microprocessor. The intelligent controller 12 receives the outputs of the various position sensors via line 166.

It is to be understood that the embodiment which will be subsequently described is but one of a number 20 of possible implementations of the position calibration method of the present invention. Other embodiments can include separate hardware for controlling each of the different positioning and sensing sequences.

Utilizing the system £>f Figure 1, the intelligent 25 controller 12 maintains ^internal set of position registers, sensor state registers, and toggle registers. For purposes of explanation, it is to be assumed that motors 14 are stepper motors such that the drive signals provided to the motors by motor driver cir30 cuitry 22 are in the form of a series of unit steps. As such, the number of unit steps provided to a particular stepper motor 14 can be related to the position of the structural member which is controlled by the particular stepper motor 14. Thus, the position 35 register maintained by intelligent controller 12 are

<img file="IL69862A_D0004.tif" />

intended to keep track of the number of unit steps supplied to each stepper motor. The toggle registers are intended to keep track of whether each of the sensors has gone through a transition in its output.

state, thereby indicating that a home position has been encountered. The sensor state registers are intended to maintain a record of the state of each position sensor at particular points in the calibration sequence.

In practice, the position registers can be up/down 10 counters such that a net-position count will be maintained. For example, when the intelligent controller 12 provides a sequence of unit steps corresponding to the rotation of the stepper motor in a particular direction, the up/down counter would increment its 15 count by the number of steps within the command. When the intelligent controller 12 provides a number of steps which will cause the motor to rotate in an opposite direction, the up/down counter will decrement its count by that number of steps. Thus, the resulting 20 count within the up/down counter will be a net position count.

Referring to Figure 10, step 168 corresponds to the power-on and reset mode of the control system. Typically, this includes the application of power to 25 the stepper motors, and the stablization of the various control circuits, drive circuits, and the various structural members within robotic arm 10. After the power-on/reset phase in step 168, step 170 is processed in which the state of the position registers is set to 30 zero. This provides an initial starting point for the intelligent controller 12. Preferably, the arm 10 is in its home position at this step. After step 170, step 172 is executed in which the robotic arm 10 is operated in its normal mode, or in an idle state.

At some time during this step 172, either as in initialization procedure or periodically during the normal operation of the arm, the intelligent controller will be instructed to enter step 174, which corresponds to the start of the homing or position calibra5 tion operation. When such an operation is desired, step 176 is first executed. In this step, the robotic arm is moved from whatever its current position back to the position which corresponds to a position register state of all zeros. Preferably, when in step 170 the position registers were set to zero, the robotic arm was at or near its home position. If such were the case when step 170 was processed, step 176 will cause the arm to move to a point which is in the approximate area of the home position of the arm 10.

During step 176, all of the structural members are driven simultaneously. Typically, step 176 is executed over a predetermined time. By examination of the position register states, the intelligent controller 12 can determine which structural member is required to move the furthest. The motor corresponding to that particular structural member would then be provided with an uninterrupted series of steps, the number of which would correspond to the number contained in the appropriate position register. The stepper motors corresponding to those structural members which are a lesser number of steps away from position zero will be provided with a drive signal in which steps are inserted periodically during the signal to the structural member having to move the furthest. The period with which these steps are inserted is a function of the number of steps which must be output to the particular motor and the amount of time allotted therefor.

After step 176 is completed, the toggle registers are set to zero in step 178. This prepares the toggle 35 registers to record whether or not a transition has occurred in each of the position sensor outputs for each of the structural members. After step 178, step 180 is executed in which the sensor state registers are set to the then current values of ±he position sensor 5 outputs. After step 180 has been completed, step 182 is executed in which all of the motors are commanded, simultaneously, to move a predetermined unit step in the direction of the transition point of each position, sensor.

As discussed in connection with the position sensors, in the preferred embodiment, the position sensors are located so that there is a single transition within the range of motion of the particular structural member with which it is associated. As such, by knowing the state of the position sensors, the direction of the state transition, and hence that of the home position, can be derived therefrom.

Thus, in step 182, the intelligent controller will be able to determine the direction in which to command 20 each motor to move, such that the resulting movement will be toward the sensor edge or transition point. The speed at which each motor is commanded to turn in step 182 is typically lower than that used in step 176. At the end of each unit of; movement in step 182, the 25 intelligent controller 12 examines the outputs of all of the position sensors, step 184. If the sensor output for a particular sensor is equal to the sensor state in its corresponding sensor state register, it is clear that no transition has occurred, and that the 30 home state for the structural member has not yet been reached. Step 184 is repeated for each of the position sensors. If, with respect to a particular position sensor, a change in state is detected, the corresponding toggle register is set to one in step 186. After all 35 position sensors have been examined in step 184 and after any transitions which are detected are accounted (B (.

for in step 186, step 188 is then processed. In step 188, the intelligent controller 12 determines whether all of the toggle registers have been set to one. If not, the system returns to step 180 in which the sensor 5 registers are reset to the then current output states of the position sensors. Step 182 is then repeated to move each of the structural members one more unit step toward their sensor edge. For the position sensors in which a transition had previously occurred, 10 the execution of step 182 will cause the structural member to move one step in the opposite direction.

Thus, when step 184 is again executed, the corresponding toggle register will again be set to one in step 186* Thus, the sequence of steps 180, 182, 184, 186 15 and 188 will move each of the structural members toward their corresponding position sensor edge, and once such edge is encountered for each structural member, cause that structural member to vary in position about that sensor edge.

After all of the toggle registers have.been set to one, the system will proceed out of step 188 and into step 190. In step 190, the intelligent controller will determine whether or not the sensor is in its designated home state. Recall that the home state is in arbitrarily selected output value, which has been selected for each of the position sensors. The value of the home state designates upon which side of the transition in the position-detector state a home state will be said to be located.

Steps 190 and 192, which are entered if the result of step 190 indicates that all toggle registers have been set to one, are executed in connection with one motor at a time. Additionally, the motors are operated upon in a predetermined sequence. In this set of step sequences, the coupling between the various structural members is taken into account to minimize the interaction of one member upon the others. Thus, the motors and their associated structural members are operated 5 upon in a sequence which begins with the structural member having the greatest amount of coupling to the other structural members, and continues with the structural member having the next lower amount of coupling. The sequence ends with the structural member 10 and associated motor having the least degree of coupling with respect to the other structural members. Thus, in the outwardly coupled embodiment described above, the upper arm 28 will be processed first followed by the forearm 32 and then the hand 36.

If in step 190 the structural member being cur־ rently operated upon is not in its home state, the intelligent controller 12 will cause the motor to move the structural member, slowly, one unit step toward the home state. A gradual or slow movement is utilized here to prevent the movement of the particular structural member from creating vibrations which could potentially cause the states of the other position sensors to change. In the preferred embodiment, because of this vibration problem, the base position is 25 modified first, even though the base 24 has a low degree of coupling. This avoids moving a lot of mass at the end of the calibration sequence.

Steps 190 and 192 are repeated for each of the structural members, in accordance with the sequence 30 discussed above, until all position sensors are in their respective home states. When this occurs, steps 194 and 196 are then executed for each motor and associated structural member and in accordance with the designated sequence, as discussed above. As can be 35 seen from a comparison of steps 190/192 and 194/196, that the operations are similar. The purpose of executing step 194 and step 196 is to provide a second check of the position sensor states and to allow for the situation where a structural member, subsequently, has been moved out of position duef to the movement of structural members processed later in the sequence in steps 190 and 192.

If, in step 194, the position sensor for the particular motor and associated structural member is in the desired home state, step 198 is then processed. In step 198, the intelligent controller 12 causes the motor to move slowly one step away from the desired home position. The sensor state is then checked in step 200. Step 200 is provided to ensure that the sensor is within one step of its transition point. Thus, if in step 200, the sensor is still in its home state, this is indicative that the transition point has not been crossed. Thus, the system will proceed back to step 198 and move the structural member and motor one step closer to the home position.

Steps 194/196 and 198/200 are repeated for each motor and associated structural member, and in accordance with the predetermined sequence discussed above. When all of the position sensors have been backed off a unit step from their home state, the system is then in a position-calibrated mode. In this mode, all of the position sensors are positioned within one unit step of the position sensor transition point. Hence, all of the structural members are in one unit step of their home position.

Upon final completion of step 200, the intelligent controller 12 proceeds to step 202. While step 202 is not essential to the operation of the method of the present invention, the execution of this step is valuable in determining Whether any positional errors exist. In step 202, the contents of the position registers are examined. Assuming that the robotic arm was in its home position when the position registers were set to zero in step 170, the position registers should theoretically equal zero at step 202. If there are non-zero quantities in the pos-ition registers, this is indicative that the robotic arm has somehow been moved by an external force, or that the stepper motors have slipped in some manner.

The quantity in the position registers is also valuable in calculating the amount of force being applied by a particular structural member to an external object. This is because force supplied by the structural member can be related to the number of steps slipped by the associated drive motor 14. This force is also a function of the moment arm associated with the structural member and with the speed of the arm.

After־step 202 has been executed, the system returns to step 170 in which the position registers are set to zero. This then causes the zero condition of the position registers to correspond to the home position of the robotic arm. After step 170, the system returns to normal operation and/or idle operation until another homing or position calibration operation is desired.

As a result of the provision of position sensor associated with each of these structural members and with the implementation of the method of position calibration described above, a robotic arm and control system is obtained which provides a semi-closed loop control. As such, there is improved performance over the prior open loop, tendon driven, robotic arm, at a cost and complexity which is far below that of the more precise and costly industrial prior art robotic arms.

Unlike the prior art industrial arms which employ shaft encoders, the present invention employs position sensors which generate a transition between two different states when the home position or reference point is reached for the particular structural member. The present invention is then able to utilize this transition information to provide ׳an accurate position 5 calibration.

This capability is further enhanced by the method of routing the tendons or drive cables, from drive capstans to idler pulleys and to drive pulleys, as described above. The utilization of a swaged ball on 10 each cable, which is then inserted into the capstan and idler pulleys, further enhances the precision of the position calibration system, as well as the precision of operation of the arm during normal operation.

The terms and expressions which have been employed 15 herein are used as terms of description and not of limitation, and there is no intention, in the use of such terms and expressions, of excluding equivalents of the features shown and described, or portions thereof, it being recognized that various modifications are 20 possible within the scope of the invention claimed.