Magneto-rheological clutch with sensors measuring electromagnetic field strength

Abstract

Systems and methods relating to a clutch system for use in controllably transmitting torque from an input shaft to an output shaft. The clutch system has a torque transmission fluid that has a viscosity that changes based on the strength of an electromagnetic field passing through the fluid. A number of sensors are placed at different radial locations on the torque transmission disks to detect the strength of the electromagnetic field. Based on the strength of the electromagnetic field, the amount of torque being transmitted from the input shaft to the output shaft can be adjusted. Also disclosed is a distributed actuation architecture that uses this clutch system. The distributed actuation architecture allows for the use of a single drive motor in conjunction with multiple instances of the clutch system to actuate a mechanical linkage, such as a robotic arm.

Claims

1. A clutch system for transmitting torque from an input shaft to an output shaft, the system comprising: a plurality of input torque transmission disks coupled to said input shaft; a plurality of output torque transmission disks coupled to said output shaft, portions of said input torque transmission disks being interleaved with portions of said output torque transmission disks; a torque transmission fluid disposed between said interleaved input torque transmission disks and output torque transmission disks, said torque transmission fluid having a viscosity which changes based on a strength of an electromagnetic field passing through said transmission fluid; field means for providing an electromagnetic field which passes through said torque transmission fluid; a plurality of sensors for measuring a strength of said electromagnetic field at different radial positions along said input torque transmission disks or along said output torque transmission disks, wherein said plurality of sensors measures said electromagnetic field in said torque transmission fluid.

2. A system according to claim 1, further comprising a control system for controlling an output of said field means based on an output of said plurality of sensors, said control system controlling a strength of said electromagnetic field to control an amount of torque being transmitted.

3. A system according to claim 1, wherein said electromagnetic field is a magnetic field.

4. A system according to claim 3, wherein said torque transmission fluid is a magneto-rheological fluid.

5. A system according to claim 1, wherein said electromagnetic field is an electrical field.

6. A system according to claim 5, wherein said torque transmission fluid is an electro-rheological fluid.

7. A system according to claim 5, wherein said field means is an electromagnet.

8. A system according to claim 7, wherein said electromagnet comprises coils using wire having a non-circular cross-section.

9. A system according to claim 7, wherein said electromagnet comprises coils using wire having a rectangular cross-section.

10. A system according to claim 7, wherein said electromagnet comprises coils using wire having a circular cross-section.

11. A system according to claim 1, wherein said system is deployed on a robot.

12. A system according to claim 1, wherein said input shaft is mechanically coupled to a drive motor, said drive motor being located at a base of a robot.

13. A system according to claim 1, wherein said system is located at a joint of a robot.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The embodiments of the present invention will now be described by reference to the following figures, in which identical reference numerals in different figures indicate identical elements and in which:

(2) FIG. 1 is a graph showing the HIC for a simulated single axis manipulator;

(3) FIG. 2 illustrates a typical MR torque transmission clutch;

(4) FIG. 3 illustrates the magnetic field lines in the clutch of FIG. 2;

(5) FIG. 4 is a diagram illustrating the various dimensions to be used in the calculations according to one aspect of the invention;

(6) FIGS. 5A and 5B illustrates the field dependence of yield stress for a specific MR fluid (FIG. 5A) and for a specific ER fluid (FIG. 5B);

(7) FIG. 6 illustrates the reflected inertia vs. actuator torque of an MRF clutch for T.sub.c=5Nm, l.sub.d=1 mm, .sub.y=40 Kpa, =7.710.sup.3 kg/m.sup.3;

(8) FIG. 7 shows the various radii used in the calculations according to one aspect of the invention;

(9) FIG. 8 is a graph showing the mass of simplified MR fluid clutch model vs. torque capacity;

(10) FIG. 9 is a graph showing the mass of simplified ER fluid clutch model vs. torque capacity;

(11) FIG. 10 is a graph showing yield stress vs. electric field of LID 3354S for varying temperature;

(12) FIG. 11 is a cross-sectional view of an MR clutch according to one aspect of the invention;

(13) FIGS. 12A and 12B are cross-section views of a section of the MR clutch of FIG. 11 showing the relationship between the input torque transmission disks and the output torque transmission disks;

(14) FIG. 13 is a side cross-sectional view of the section of the MR clutch of FIG. 12A showing the spacing between the input and output torque transmission disks;

(15) FIG. 14 is a side cross-section view of the MR clutch of FIG. 13 showing the placement of a sensor for determining the intensity or strength of an electromagnetic field passing through the torque transmission fluid;

(16) FIG. 15 is a front view of some of the disks of the MR clutch of FIG. 11 illustrating the radial placement of the various sensors in an output torque transmission disk;

(17) FIG. 16 is a side cross-sectional view of the components in FIG. 15 illustrating the placement of the temperature sensor, the electromagnetic sensor, and the voids where dowels may be placed to mechanically couple the disks together;

(18) FIG. 17 is a block diagram illustrating torque control using the Hall sensors in the implementation of the invention illustrated in FIG. 11;

(19) FIG. 18 is a schematic diagram of a distributed active semi-active actuation system according to another aspect of the invention;

(20) FIG. 19 is a schematic diagram of a pair of clutches in an antagonistic distributed active semi-active configuration;

(21) FIG. 20 is a cut-away view of an antagonistic distributed active semi-active actuation system for use as a bidirectional actuation system using a single driving motor;

(22) FIG. 21 is a schematic diagram of a system that uses multiple instances of the bidirectional actuation system of FIG. 20 using a single driving motor to drive the various instances of the actuation system;

(23) FIG. 22 is an illustration of robotic arm with the drive motor located at a bottom portion of the robot;

(24) FIG. 23 is a front view of the robotic arm of FIG. 22; and

(25) FIG. 24 is a cross-sectional cut-away view of the robotic arm of FIG. 22 illustrating the four clutch mechanisms used.

DETAILED DESCRIPTION OF THE INVENTION

(26) Magneto-rheological (MR) and electro-rheological (ER) fluids are a suspension of micrometer-sized particles in a carrier fluid. When subjected to an externally applied fielda magnetic field for MR fluids, and an electric field for ER fluids, the particles aggregate into columns aligning themselves in the direction of the applied field. Subsequently, the columns act to resist shearing of the fluid perpendicular to the field. The apparent yield stress of the fluid is dependent on, and increases with the intensity or strength of the applied field.

(27) Using either MR or ER fluids, a clutch can be constructed such that the amount of torque it transmits can be controlled by the intensity or strength of an applied field. FIG. 2 is a cross-section of a multi-disk style MR fluid clutch while FIG. 3 illustrates the flux path within the MR fluid clutch. The input shaft breaks out into a set of input disks which are aligned in parallel to a set of output disks attached to the output shaft. MR fluid fills the volume between input and output disks. Rotation of the input shaft causes shearing in the fluid with respect to the output shaft. By energizing the electromagnetic coil, a field is induced in the MR fluid altering its apparent viscosity. The outer casing of the MR clutch acts as the magnetic flux path required to complete the magnetic circuit. An ER clutch is formed much the same way, however does not require a magnetic coil. Applying a high voltage to the input shaft while grounding the output shaft allows an electric field to be induced in the fluid between the input and output disks. Again, this field controls the apparent viscosity in the fluid.

(28) A number of models have been developed to describe the behavior of both MR and ER fluids. The Bingham visco-plastic model is commonly used to represent the sheer stress of the fluid as a function of the applied field, and shear rate. The model is given by

(29) $\begin{array}{cc}={}_y\left(\right)+\frac{}{z},>{}_y& \left(3\right)\end{array}$
where is the shear stress, .sub.y is the field dependent yield stress, is the applied field (magnetic field intensity H in an MR device or electric field E in an ER device), is the Newtonian viscosity, and dv/dz is the velocity gradient in the direction of the field. Applying the Bingham visco-plastic model to a clutch, we can define r as the radius from the rotational axis, and l.sub.f as the thickness of the fluid filled gap between input and output disks. In situations where r>>l.sub.f for r[R1,R2] (refer to FIG. 4), the velocity gradient becomes constant. We can then rewrite eqn. (3) as
=.sub.y()+{dot over ()}(r), >.sub.y(4) where the shear rate is defined as

(30) $\begin{array}{cc}\stackrel{.}{}=\frac{r}{l_f}& \left(5\right)\end{array}$
and is the angular velocity between input and output shafts of the clutch. The torque produced by a circumferential element at a radius r is given by.
dT=2r.sup.2dr.(6)

(31) We define a clutch as having N output disks. Substituting (4) into (6) and integrating across both faces of each output disk, we arrive at

(32) $\begin{array}{cc}\begin{array}{c}T=2N{}_{R_1}^{R_2}2\left[{}_y\left(\right)r^2+\frac{r^3}{l_f}\right]r\\ =4N\left[\frac{{}_y\left(\right)\left(R_2^3-R_1^3\right)}{3}+\frac{\left(R_2^4-R_1^4\right)}{4l_f}\right]\end{array}& \left(7\right)\end{array}$
as the torque transmitted by an N-disk clutch. Data relating the yield stress .sub.y of a fluid to an applied field are generally published by the manufacturer. Data for two commercially available fluids are reported in FIGS. 5A and B. FIG. 5A is the field dependence of yield stress for the MR fluid MRF-132DG produced by Lord Corp. while FIG. 5B is the field dependence for yield stress for the ER fluid LID 3354S produced by Smart Technology Ltd. The viscosity of the carrier fluid is typically in the range of 0.1 to 0.3 Pa-s. The maximum torque transmission capability of an MR/ER clutch is dependent on the maximum yield stress the material can produce. MR fluids exhibit saturation in their yield stress at high field strengths. This is a result of the underlying physics, and limits the amount of torque a particular MR fluid can transmit in clutch applications. ER fluids on the other hand, experience electrical breakdown at high field strengths, also limiting their torque transmission capacity. MR fluids can produce maximum yield stresses typically in the range of 50 to 100 kPa depending on their chemistry. Alternatively, this value for ER fluids does not typically exceed 10 kPa. It should be noted that more recent research into the field of ER fluids has produced materials exhibiting a yield stress on par and even exceeding that of MR fluids.

(33) Both MR and ER fluids respond to an applied field on the order of 1 ms. However, in the case of MR clutch, actuation response becomes delayed due to field propagation through the magnetic circuit. As the magnetic field propagates from the coil, it is met by an opposing field produced by induction currents in the magnetic circuit. The response delay produced by this effect is exacerbated by the use of conductive material in the magnetic circuit. Special design techniques not unlike those used to construct large power transformers can be adapted to minimize the magnitude of the induced currents.

(34) MR/ER actuators have the characteristic of replacing the reflected rotor inertia of the motor with the reflected inertia of the clutch output shaft and disks. The benefit of MR/ER actuators is their high torque to output moment of inertia ratio as compared to servo motors. To show this, we approximate the radius of the output shaft to be equivalent to R.sub.1. The moment of inertia of a single output disk, J.sub.d is given by

(35) $\begin{array}{cc}{J}_d=\frac{1}{2}{}_d{l}_f\left(R_2^4-R_1^4\right)& \left(8\right)\end{array}$
where .sub.d is the mass density of the disk material, l.sub.f is the thickness of the disk (commonly between 0.5 to 1 mm), and R.sub.1 and R.sub.2 define the minor and major radii respectively, of the output disk. If we consider the torque transmitted solely by the field dependent yield stress of the MR/ER fluid, the torque transmission of a single disk is then given by,

(36) $\begin{array}{cc}{T}_d=\frac{4}{3}{}_y\left(R_2^3-R_1^3\right)& \left(9\right)\end{array}$

(37) Furthermore, if we consider R.sub.1 to be small, that is R.sub.2>>R1, then the contribution of the shaft region to both (8), and (9) is also small. By allowing R1 to equal zero, we can approximate the torque-inertia ratio of a single disk to be

(38) $\begin{array}{cc}=\frac{T_d}{J_d}=\frac{8}{3}\frac{{}_y}{{}_d{l}_d{R}_2}& \left(10\right)\end{array}$

(39) As observed the ratio becomes less favorable as R.sub.2 increases. This however is not the final measure that dictates the actuator's suitability. To grasp the overall effects of increasing radius, and hence torque capacity, the reflected inertia at the joint should be considered. The reason for this is that as radius increases along with torque capacity, the gear ratio required to amplify the actuator's torque decreases. As the actuator inertia multiplies the square of the gear ratio to arrive at the reflected inertia at the joint, the analysis becomes important. The reflected inertia of the MR/ER clutch at the manipulator joint is given by

(40) $\begin{array}{cc}{J}_c^{}=\frac{1}{2}{}_d{l}_{d}N\left[R_2^4-R_1^4\right]G_r^2& \left(11\right)\end{array}$
where we have included N to multiply the inertia by the number of disks in the clutch. The gear ratio G.sub.r is defined as

(41) $\begin{array}{cc}{G}_r=\frac{T_c^{}}{T_c}& \left(12\right)\end{array}$
where T.sub.c is the desired torque at the joint, and T.sub.c is the output torque of the clutch. Rearranging (9) to show the outer radius R.sub.2 as a function of the clutch output torque yields

(42) 0$\begin{array}{cc}{R}_2={\left(\frac{3}{4}\frac{T_c}{{}_{y}N}+R_1^3\right)}^{1/3}& \left(13\right)\end{array}$

(43) We can then write the equation representing the reflected inertia of an MR/ER clutch at the manipulator joint as a function of the clutch torque.

(44) $\begin{array}{cc}{J}_c^{}=\frac{1}{2}{}_d{l}_{d}N\left[{\left(\frac{3}{4}\frac{T_c}{{}_{y}N}+R_1^3\right)}^{4/3}-R_1^4\right]{\left(\frac{T_c^{}}{T_c}\right)}^2& \left(14\right)\end{array}$

(45) FIG. 6 shows the values of reflected actuator inertia for a typical MR clutch having between one and 20 disks. The plot also includes equivalent values for commercially available low-inertia servo motors.

(46) To establish the order in which clutch mass compares to that of servo motors as well as the rate at which clutch mass increases with respect to transmittable torque capacity, a mass-torque relationship has to be developed. This is developed using simplified geometric models but the conclusions are still valid.

(47) To develop a relationship between clutch mass and torque capacity for MR fluid clutches, we consider the simplified geometric model detailed in FIG. 7. FIG. 7 is a cross-sectional view of a simplified geometric model of an MR fluid clutch. In the figure, a single amalgamated shaft extends the length of the clutch. Input and output disks are defined as having equivalent geometric and material properties, sharing minor radius R.sub.1 and major radius R.sub.2. The electromagnetic coil windings are contained between R.sub.2 and R.sub.3, and R.sub.4 defines the outer surface of the ferrous core used to complete the magnetic circuit.

(48) Based on the configuration in FIG. 7, we will solve for required parametric values through the application of magnetic circuit analysis. We divide the reluctance of the core Rc into three sections, namely R.sub.c1, R.sub.c2, and R.sub.c3. The symmetric geometry of the model dictates the reluctance R.sub.c2 to be equivalent to that of R.sub.c3. Thus we define the reluctance of the core to be
.sub.c=.sub.c.sub.12.sub.c.sub.23(15)
where R.sub.c23=R.sub.c2=R.sub.c3. We have defined a clutch by the number of output disks N coupled to the output shaft. For N output disks, a clutch is required to have N1 input disks, and a total of 2N MR fluid interface gaps positioned between the input and output disks. In the simplified model of FIG. 7, we define both geometric, and material properties of the input and output disks to be identical. The disk pack assembly thus contains 2N1 disks, and 2N MR fluid interface gaps. The reluctance of the disk pack assembly R.sub.p can then be written as
.sub.p=(2N1).sub.d+2N.sub.f(16)
where R.sub.d, and R.sub.f are the reluctance of a single disk, and the reluctance of a single MR fluid interface gap, respectively. The reluctance of a material is defined by

(49) $\begin{array}{cc}\mathrm{.Math.}=\frac{l}{{}_0{}_{r}A}& \left(17\right)\end{array}$
where l is the mean length of the flux path through the material, .sub.0=410.sup.7 H/m is the permeability of free space, .sub.r is the relative permeability of the material, and A is the cross sectional area of the material perpendicular to the flux path. Assuming the mean flux path through any of the circuit members lies at its geometric center, we can then derive the reluctance of the individual components of the simplified clutch model to be

(50) $\begin{array}{cc}{\mathrm{.Math.}}_{c_1}=\frac{l_p+l_c}{{}_0{}_{r_s}\left(R_4^2-R_3^2\right)}& \left(18\right)\\ \begin{array}{c}{\mathrm{.Math.}}_{c_{23}}={}_{\frac{R_2+R_1}{2}}^{\frac{R_4+R_3}{2}}\frac{r}{{}_0{}_{r_s}\left(2r\right)l_c}\\ =\frac{\ln \left(\frac{R_4+R_3}{R_2+R_1}\right)}{2{}_0{}_{r_s}{l}_c}\end{array}& \left(19\right)\\ {\mathrm{.Math.}}_d=\frac{l_d}{{}_0{}_{r_s}\left(R_2^2-R_1^2\right)}& \left(20\right)\\ {\mathrm{.Math.}}_f=\frac{l_f}{{}_0{}_{r_f}\left(R_2^2-R_1^2\right)}& \left(21\right)\end{array}$

(51) Here, .sub.rs is the permeability of steel, the material used for both the core and disks, .sub.rf is the permeability of the MR fluid, W.sub.d is the thickness of a single disk, l.sub.f is the distance between input and output disks forming the MR fluid gap, l.sub.c is the thickness of the equivalent core sections, and l.sub.p is the length of the disk pack, given by
l.sub.p=(2N1)l.sub.d+2Nl.sub.f(22)

(52) The flux in the circuit is then given by

(53) $\begin{array}{cc}=\frac{I}{{\mathrm{.Math.}}_c+{\mathrm{.Math.}}_p}& \left(23\right)\end{array}$
where I is the total electric current through the cross section of the magnetic coil defined as,
I=l.sub.p(R.sub.3R.sub.2)J.sub.w(24)

(54) In eqn. (24), J.sub.w is the current density of the coil cross section. The magnetic field intensity H at any point within the circuit is related to the circuit flux by

(55) $\begin{array}{cc}H=\frac{}{{}_0{}_{r}A}& \left(25\right)\end{array}$
where again, .sub.r and A are respectively the relative permeability, and cross sectional area of the material at which the magnetic field intensity H is to be determined. We now define the parameter *.sub.y as the maximum yield stress at which the MR fluid is to operate. Using data provided by the MR fluid manufacturer (see FIG. 5A) relating the yield stress of the fluid to the applied magnetic field, we define H* as the magnetic field intensity in the MR fluid required to produce the yield stress *.sub.y. Rearranging eqn. (25), and substituting the appropriate MR fluid geometric and material values, we define * as the flux required in the circuit to produce H* in the MR fluid,
*=.sub.0.sub..sub.f(R.sub.2.sup.2R.sub.1.sup.2)H*(26)
where R.sub.2 is uniquely defined by the parameters T.sub.c, N, R.sub.1 and *.sub.y (refer to eqn. (13)). Thus, for the given set of fixed parameters given in Table I, we solve for the values of R.sub.3, R.sub.4 and l.sub.c that satisfy eqn. (23) for =*, while simultaneously minimizing the clutch mass m.sub.MRC, given by
m.sub.MRC=m.sub.c+m.sub.p+m.sub.s+m.sub.w
m.sub.c=[(R.sub.4.sup.2R.sub.3.sup.2)l.sub.p+2(R.sub.4.sup.2R.sub.1.sup.2)l.sub.c].sub.s
m.sub.p=(2N1)(R.sub.2.sup.2R.sub.1.sup.2)l.sub.d.sub.s+2N(R.sub.2.sup.2R.sub.1.sup.2)l.sub.f.sub.f
m.sub.w=(R.sub.3.sup.2R.sub.2.sup.2)l.sub.p.sub.cu
m.sub.s=R.sub.1.sup.2(l.sub.p+2l.sub.c).sub.al,(27)
where m.sub.c is the mass of the core, m.sub.p is the mass of the disk pack assembly which includes the MR fluid, m.sub.s is the mass of the shaft, and m.sub.w is the mass of the magnetic coil. In eqn. (27), .sub.s, .sub.f, .sub.cu, and .sub.al are respectively the mass densities of steel, MR fluid, copper, and aluminum. FIG. 8 shows the mass-torque relationship of the simplified MR clutch model and compares it to a commercially available servo motor. It is noted that due to the mass overhead associated with the material required to form the magnetic circuit, the mass-torque ratio of MR fluid clutches is less favorable at very low values N.

(56) TABLE-US-00001 TABLE I PARAMETER VALUES FOR SIMPLIFIED MR CLUTCH MODEL l.sub.d = 1 10.sup.3 m l.sub.f = 5 10.sup.4 m R.sub.1 = 1 10.sup.2 m T.sub.y.sup.* = 40 kpa J.sub.w = 2.5 10.sup.6 A/m.sup.2

(57) The mass of an ER fluid clutch can be similarly approximated by again considering the simplified geometric model of FIG. 7. By assigning R.sub.3 the value of R.sub.2, the electromagnetic coil effectively vanishes from the model. We can then express the mass of an ER fluid clutch in terms of the mass equations previously developed, that is
m.sub.ERC=m.sub.p+m.sub.s+m.sub.c(28)
however, we substitute the mass density of aluminum .sub.al for all components. As ER fluid clutches do not require ferromagnetic metals, there is more freedom to select lighter materials. Again, R.sub.2 is defined by eqn. (13). Thus, given the values of the clutch parameters in Table II, we develop the mass-torque relationship represented in FIG. 9.

(58) In the developed mass-torque relationships of both MR and ER fluid clutches, we observe superior characteristics over the commercially available servo motor. More impressive is the fact that the mass-torque ratio improves dramatically as the torque transmission capacity of the clutch is increased.

(59) TABLE-US-00002 TABLE II PARAMETER VALUES FOR SIMPLIFIED ER CLUTCH MODEL l.sub.d = 1 10.sup.3 m l.sub.f = 5 10.sup.4 m l.sub.c = 7.5 10.sup.3 m R.sub.4 = R.sub.2 + l.sub.c R.sub.1 = 1 10.sup.2 m T.sub.y = 5 kpa

(60) One possible issue with both the MR and ER actuators is that they both exhibit hysteric behavior. With MR actuators, since they are constructed using ferromagnetic materials to complete the magnetic circuit, these materials exhibit hysteresis in its magnetic induction B to an increasing and decreasing applied magnetic field H. This means a hysteric relationship between the input current and the output torque transmission of an MR clutch.

(61) ER actuators can also display hysteric behavior as well. However, for ER actuators, it is the ER fluid and not the surrounding material that displays the field dependent hysteric behavior.

(62) The properties of ER fluids are known to change with temperature. Particularly, the yield stress of ER fluids varies with respect to temperature (FIG. 10). The integration of temperature sensors into an ER clutch would thus prove useful in mitigating error.

(63) Referring to FIG. 11, a cross-sectional view of an MR clutch developed based on the above noted calculations is illustrated. The MR clutch has an input shaft that is belt driven which, in turn, rotates the input torque transmission disks (green disks in the Figure) about the central axis of the clutch. The input torque transmission disks are interleaved with the output torque transmission disks (blue disks in the Figure) and between these disks is a torque transmission fluid. In the case of the MR clutch, the torque transmission fluid is an MR fluid. The magnetic coil required to create the magnetic field passing through the MR fluid is shown as the orange/copper colored component in the Figure. With the magnetic coil energized, the magnetic field passes through the MR fluid and a rotation of the input torque transmission disks about the central axis of the system causes the output torque transmission disks to similarly rotate about the central axis.

(64) The mechanical operation of the clutch system can best be explained with relation to FIGS. 12A and 12B. Both Figures are isometric cross-sectional views of a portion of the clutch system. In both Figures, it can be seen that the output torque transmission disks (in blue) are interleaved with the input torque transmission disks (in green). The orange/copper colored electromagnetic coil is clearly visible in both Figures. The outer housing (colored pink) has a top groove to accommodate an input drive belt. FIG. 13 is a side cross-sectional view of a portion of the clutch system. The top portion of the outer housing (in pink) is where the drivebelt is positioned. The torque transmission fluid is in between the input torque transmission disks (in green) and the output torque transmission disks (in blue).

(65) Referring to FIG. 14, a side cross-sectional view of a portion of the clutch system is illustrated. To determine the strength or intensity of the electromagnetic field passing through the torque transmission fluid, sensors 100 are placed in at least one of the output torque transmission disks. In FIG. 14, the sensor 100 is shown as the white rectangle between two blue torque output transmission disks. The PCB (printed circuit board) for this sensor 100 is shown as the green rectangle connected to the sensor by a thin white line. It should be noted that while the sensor 100 is placed with the output torque transmission disks in this embodiment of the invention, the sensor 100 may equally be placed with the input torque transmission disks.

(66) To determine the strength or intensity gradient of the electromagnetic field, multiple sensors 100 may be placed at different locations on either of the input or the output torque transmission disks. This may be done by placing sensors 100 at different radial distances from the central axis. Such a placement would provide a better reading as to the intensity or strength of the electromagnetic field through the torque transmission fluid.

(67) It should be noted that for an MR clutch, the sensor 100 could be Hall effect sensors that measure the intensity of magnetic fields. For an ER clutch, the sensor 100 could measure the intensity of the electric field through the torque transmission fluid.

(68) To better explain the placement of the sensors 100, FIG. 15 is a front view of a portion of the clutch showing one of the output torque transmission disks and showing the placement of the sensors 100 in suitably dimensioned cavities on this disk. The sensors 100 (the white quasi-rectangular components in the Figure that are attached by wires to green PCBs) are placed so that they are all at different radial distances from the central axis of the clutch. The output of these sensors 100 may be used by a control system to accordingly adjust the strength or intensity of the electromagnetic field. As mentioned above, the higher the intensity or strength of the electromagnetic field, the higher the viscosity of the torque transmission fluid and, accordingly, the greater the amount of torque being transmitted from the input disks to the output disks.

(69) Regarding the sensors for sensing the electromagnetic field intensity, for the MR case, ideally, the magnitude of the magnetic field would be linearly proportional to the input, or drive current. However, due to the ferromagnetic components required for the magnetic circuit, a hysteric relationship is developed between input current and magnitude of the magnetic field in the MRC (MR clutch). For this reason, an MRC explicitly designed for those applications would be better served by bipolar Hall sensors to provide the feedback signal to the controller. Bipolar capability is preferred in the Hall sensors for proper control due to the potential for overshoot when attempting to drive the magnetic field to zero as well as correcting for the hysteric characteristic behavior.

(70) It follows that the location of the sensors 100 should reflect the parameter, or value they are intended to measure, namely the magnitude of the magnetic field in the fluid for the MR clutch. Theoretically, the magnetic flux can be measured anywhere in the magnetic circuit, however, unaccounted dynamics, as well as other circuit anomalies could force us to revisit the problems of sensor collocation. In one implementation, four Hall sensors are thus positioned radially at the axial center of the clutch pack such that the orientation of the Hall element is normal to the magnetic flux.

(71) While the fluid itself reacts on the order of milliseconds, the dynamics of the magnetic field in the fluid can be appreciably slower. This reflects the natural inductance developed by the ferromagnetic material of the clutch. As a result, a radial gradient can form in the magnetic field as it propagates outwards radially. Due to the fast dynamics of the fluid, the result is a variable radial distribution of the shear stress developed in the MRF and applied to the disks' surface during dynamic input current activity. To more accurately calculate output torque during such dynamic activity, Hall sensors are positioned at incremental radii covering the shear region within the clutch pack.

(72) It should be noted that the clutch system illustrated in the Figures is an MR clutch system. However, with some adjustments in the materials, the clutch system can similarly be constructed as an ER clutch.

(73) Regarding the manufacture of the MR clutch system illustrated in the Figures, the different materials are color-coded in FIGS. 11-14 as follows: PinkNonmagnetic material (aluminum in one embodiment of the invention) which has a mechanical connection to the outer disks, as well as being driven by the drive belt. This is sometimes referred to as the input shaft or input hub. GreenOuter disks made of ferromagnetic material (in one embodiment, steel was used). Note that these are the input disks as they are mechanically coupled to the drive belt which forms the mechanical input. RedOuter spacers. These are used to ensure disks do not come into contact with each other. They must be made of nonmagnetic material. In one embodiment, aluminum was used. BlueInner disks (ferromagnetic, steel). Note that these are the output disks as they are mechanically coupled to the output shaft located at the radial center of the clutch. YellowInner spacers (nonmagnetic, aluminum). BlackInner sealing cylinder. This member must be nonmagnetic and functions to seal the MR fluid inside the clutch pack. It is perforated by several slits which allow sensor access to the clutch pack. In one implementation, this was constructed from plastic. YellowInner spacers (nonmagnetic, aluminum). Orange/CopperThis is the electromagnetic coil. It is wound using 17 gauge square cross-section copper wire. However, other suitable materials and configurations may be used. A square cross-section copper wire was selected as it increased the density of the wire in the coil compared to regular circular cross-section copper wire. However, other configurations of wire (whether having a circular or non-circular cross-section) may be used.

(74) Given the susceptibility of torque transmission fluid to change in characteristics due to temperature changes, temperature sensors in the clutch system are recommended. Referring to FIG. 16, the electromagnetic sensors can be seen on the torque output transmission disk. Between the two sensors shown in the Figure is a temperature sensor. In FIG. 15, the temperature sensor can be seen between two electromagnetic sensors in the lower right portion of the output torque transmission disk. The temperature sensor can be used to ensure that the system operates within the operational limits of the components. Multiple temperature sensors can be placed at various positions in the system.

(75) To facilitate manufacturability without dramatically affecting the optimization of the clutch parameters, electromagnetic sensors (Hall sensors for the MR clutch) sensors are housed within a central disk sandwiched between two inner shear disks. All disks have equivalent thicknesses so that they may be cut, or stamped from a single sheet. Following a similar pattern, three temperature sensors are preferably located within the inner spacers. This provides good thermal coupling to the fluid. It is important to monitor the temperature of the fluid to ensure it does not exceed the prescribed operational limits.

(76) Referring again to FIGS. 16 and 15, the various disks used in the clutch system are held together by dowels or other suitable fastening means. The holes or voids (easily visible on the face of the various disks in FIGS. 15 and 16) are used to mechanically couple the disks together. In one implementation, dowels impale the various disks through the holes and then terminate into bores machined into the side plates. Dowels, however, are not the only option. Splines and spot welds (with attendant changes to the configuration of the various disks) may also be used.

(77) For the driving motor (the input to the system), the system illustrated uses a drive belt that travels by the channel visible at the top of the cross-sectional diagram in FIG. 14. Other drive transmission means may also be used. These include flat belts, V-belts, timing belts, roller chains, and shaft drives. Of course, use of any of these would necessitate design changes to accommodate the particular method being used.

(78) The system described above may be used to implement a clutch system with control being based on an estimation of the field sensed by the sensors. For the MR clutch, the Hall sensors may be used to sense the magnetic field and, based on an estimated value for the transmission torque (based on the sensed magnetic field intensity), the magnetic field intensity can be adjusted accordingly.

(79) In the specific MR clutch described above, the magnetic field is read at four radial locations (we can refer to these as r.sub.1 through r.sub.4) via the Hall sensors. It should be noted that any number of Hall sensors can be used. However, to detect time varying radial gradients, at least two sensors would be desirable. Given the value returned by the Hall sensors a reasonable estimate for the transmission torque can be calculated. The desired output torque is entered into the system through the torque command. The estimated output torque is subtracted from the torque command to develop the error signal. The error signal is the input to a controller (perhaps a proportional-integral-derivative (PID) controller). The output of the controller subsequently commands the MR clutch drive current (the current in the coil producing the magnetic field). If a PID controller is used in this configuration, the PID controller will drive its output to a state such that it minimizes the magnitude of the error signal. In effect, this action causes the system to deliver an estimate for torque which approaches the torque command. FIG. 17 illustrates a control system which uses this scheme.

(80) The MR clutch described above may be used in any number of ways. The clutch may be used in conjunction with suitable supporting components in a distributed active semi-active actuation configuration. The Distributed Active Semi-Active (DASA) actuation configuration locates a driving motor (the active actuator) at the base of the robot, and a semi-active device (either an MR or ER fluid clutch) at the joint (refer to FIG. 18). The gear ratios G1 and G2 are balanced to give the desired mass, and reflected output inertia at the link. Reducing G1 reduces the requirements of the clutch transmission torque which thus reduces the mass of the clutch, however the reflected output inertia is inevitably increased as G2 must then be increased to compensate. The simplified models explained above have shown how actuating a joint via an MR or ER fluid clutch can be accomplished at greatly reduced mass and reflected output inertia versus conventional servo motors. The impact on safety is immediately appreciated as the effective inertia of the link is instantly reduced. This not only improves manipulator performance, but further allows a manipulator to operate at higher velocities while maintaining safe HIC values in the event of an uncontrolled collision. Moreover, the clutch itself is back driveable, and can be thought of as exhibiting the properties of an ideal torque source. The characteristic of back drivability has been identified as more than desirable in operations requiring physical interaction between man and machine as it facilitates impedance control. While motors themselves are also intrinsically back drivable, the high ratio gear reductions they require are often not. Thus, highly performing low weight robots which implement low mass motors at the expense of high ratio gear reductions rely on torque sensors in the control loop to electronically implement back-drivable behavior. MR and ER clutches possesses a superior mass-torque ratio over their servo motor counterparts and thus can be designed to require much lower ratio gear reductions, if not developed as single unit direct-drive component, either way retaining their intrinsic back-drivability. MR and ER clutches have the added benefit of uniform torque transmission independent of armature position, unlike servo motors which suffer from nonlinearities such as cogging torque.

(81) Relocating the driving motor to the base of a robot to reduce the mass at the link in has previously had some drawbacks. It has been a restrictive practice as the newly required transmission responsible for bringing mechanical power from the base to the joint has commonly introduced unwanted friction and compliance which have reduced performance, and complicated the control system. The DASA implementation however can be controlled to operate in a region in which torque transmission is relatively immune to perturbation in the relative angular velocity within the clutch, effectively allowing the clutch to act as a mechanical power filter. This characteristic which will be explained momentarily allows the DASA system to function with less than ideal mechanical transmission while maintaining the performance and characteristics of a stiff transmission at the joint. To explain this, we consider that the Bingham model is accurate for describing the rheology of the fluid for shear stress above the field dependant yield stress .sub.y as expressed in (3). It is this Bingham region in which we wish the clutch to operate in order to benefit from the aforementioned characteristics. Below the yield stress .sub.y however, the fluid exhibits Newtonian characteristics, that is to say that grows with a non-negligible proportionality to the shear rate {dot over ( )}. We can thus attribute a field dependant shear rate threshold {dot over ( )}* below which the fluid exhibits Newtonian characteristics, and above which the Bingham model applies. To maintain the clutch in the Bingham region, the fluid at any radius r within the clutch must maintain a shear rate {dot over ( )} above {dot over ( )}*. To guarantee this condition, we define the field dependant angular velocity *, the threshold above which operation in the Bingham region is ensured as

(82) $\begin{array}{cc}{}^{*}=\frac{{\stackrel{.}{}}^{*}{l}_f}{R_1}.& \left(29\right)\end{array}$

(83) We come to (29) by rearranging (5) and substituting r with its minimum value R.sub.1. R.sub.1 must be used as it is the critical radius at which the lowest shear rate {dot over ( )} occurs within the clutch. The control strategy should therefore attempt to avoid entering the Newtonian region by controlling the motor angular velocity .sub.m to satisfy the condition
|.sub.m|=|.sub.j*|+*(30)
where .sub.j is the angular velocity of the joint, and * is a field dependent error margin selected to ensure that the dynamics of the motor have enough time to react to quickly varying values of .sub.j. * must be large enough to ensure * under all dynamic situations, however exact calculation of * may be difficult as there is a reliance on empirical data associated with the dynamics of the joint/link. Care must be taken, however, to avoid unnecessary power dissipation, which for a clutch is defined as P.sub.d=T. Because tracks *+*, the value selected for * cannot be arbitrarily large. Crossing into the Newtonian region is required to alter the direction of the torque transmitted to the link when utilizing a single clutch to implement the DASA system. As the motor must change the direction of its output rotation, the clutch torque transmission momentarily enters a dead-zone (courtesy of the Newtonian characteristics), the extent of time in which spent is dependent on the dynamic capabilities of the motor. This has the potential, and in fact the probability of creating substantial backlash, which could severely handicap the system's ability to effectively accomplish precision position control.

(84) Another configuration of the DASA system is the antagonistic configuration. An antagonistic configuration of the DASA system is intended to increase performance, and rectify the shortcomings of the single clutch DASA system discussed above. FIG. 19 is a schematic representation of the DASA antagonistic clutch configuration. FIG. 20 shows how an antagonistic clutch assembly (an MR-type clutch assembly in this case) could be coupled to a link.

(85) In the configuration in FIG. 20, the transmission shaft drives the input to both clutches through the transmission means (in this case, gears), however in opposite directions with respect to one another. The antagonistic output of the two clutches is coupled to the link. By energizing one of the two clutches, torque can be transmitted in either the clockwise, or counterclockwise direction. Thus, the antagonistic configuration allows for torque transmission to the joint to alter direction without altering the direction of the motor output, thereby eliminating the backlash introduced by the single clutch DASA. Such devices have been developed for ER fluids. Maintaining rotation of the motor shaft, the bandwidth of the antagonistic-DASA output is limited by the charging and the discharging of the relevant field required to activate the clutch pair. It is anticipated that an antagonistic DASA implementation utilizing ER clutches will have a higher controllable bandwidth over an MR DASA equivalent. This is due to the short charge time associated with electric fields as compared to the propagation of the magnetic field through the magnetic circuit of an MR clutch.

(86) If we label the two clutches of an Antagonistic DASA assembly as C1, and C2, then the motor's angular velocity should track
.sub.m=max{|.sub.j*.sub.1|,|*.sub.2.sub.j|}+*(31)
to avoid entering the Newtonian region of operation in either clutch. *.sub.1 is the angular velocity of the Bingham region threshold for clutch C1, while *.sub.2 is the angular velocity of the Bingham region threshold for clutch C.sub.2. Note that in our convention, clutch C.sub.2 has its input reversed in direction with respect to clutch C.sub.1, that is
.sub.1=.sub.j.sub.m(32)
.sub.2=.sub.j.sub.m.(33)

(87) The torque production for an Antagonistic-DASA system operating in the Bingham region is then given by

(88) $\begin{array}{cc}{T}_{A-\mathrm{DASA}}=\left[T_1\left({}_1\right)+T_2\left({}_2\right)-\frac{2.Math.{}_j.Math.}{l_f}\left(R_2^4-R_1^4\right)\right]& \left(34\right)\end{array}$
where T.sub.1 and T.sub.2 are the field dependant torques produced by clutches C1, and C2 respectively, given by

(89) $\begin{array}{cc}{T}_1=\frac{4}{3}{}_y\left({}_1\right)\left(R_2^3-R_1^3\right)\mathrm{sgn}\left({}_1\right)& \left(35\right)\\ T_2=\frac{4}{3}{}_y\left({}_2\right)\left(R_2^3-R_1^3\right)\mathrm{sgn}\left({}_2\right)& \left(36\right)\end{array}$
in which 1, and 2 are the fields produced in clutches C1, and C2, respectively. Note that the individual viscous torque contributions of C1 and C2 negate each other at the joint when .sub.j=0. As viscosity of this class of fluids does not always obey ideal models, the antagonistic configuration can effectively mitigate some nonlinearities which would otherwise have to be compensated for by the controller.

(90) The Antagonistic-DASA configuration translates a unidirectional active drive into a bidirectional actuation unit. We can consider the unidirectional drive as an independently controlled source of mechanical power. The antagonistic clutch assembly then loads the mechanical drive with the torque requirement dictated by the controller. Increasing the output power specification of the active drive allows for the introduction of multiple antagonistic clutch assemblies to independently load the active drive simultaneously. Pluralization of the antagonistic clutch assembly allows for the actuation system of multiple joints (or conceivably an entire manipulator) to be actively driven by a single motor located at the base. FIG. 21 is a schematic representation of the PA-DASA system. The motivation to pluralize the DASA system stems from the excessive cost resulting from associating both motor drive, and antagonistic clutch assembly with each joint utilizing the system. Moreover, if we attribute any credit to the anticipation that clutches of this type can be produced in mass at a considerable cost saving over servo motors, we may then expect overall production costs of PA-DASA manipulators to be much lower than equivalent manipulators actuated entirely by servo motors.

(91) Referring to FIGS. 22, 23, and 24, illustrations of the robot noted above are presented. As can be seen, the robot has a single drive motor located at the base portion of the robot (seen more clearly in FIG. 23). FIG. 24 illustrates that there are three sets of clutches in the robot, two of which are in colora first set closest to the drive motor, a second set above the first set, and a third, smaller set at the top of the robot to which only two drive belts are coupled. The third set of clutches is not shown in cut-away in FIG. 24. The drive motor shaft turns in one direction and, as in the configuration in FIG. 20, this turns two gears and thereby two drive belts in opposite directions (see FIG. 23 and the two white colored drive belts). Each drive belt then turns an input shaft to a clutch assembly in the first set of clutches. That clutch assembly in the first set then transfers torque and drives a clutch output drive belt to another clutch assembly in the second set of clutches. The second set of clutches then transfers torque further to the third set of clutches which directly moves the robot arm up or down. From FIG. 23, the two outer drive belts (dark colored) are the input drive belts to the input shafts of the two clutch assemblies in the second set of clutches. The two inner drive belts in FIG. 23 are the output clutch drive belts of the second set of clutches and are coupled to the output shafts of the clutch assemblies in the second set of clutches. These inner drive belts are then coupled to the input shafts of the third set of clutches at the top of the robot.

(92) It should be noted that the robot is capable of being rotated about its vertical axis as if on a platform. Referring to FIG. 22, a yellow cable (also visible in FIG. 23) is coupled to the first set of clutches. The yellow cable (preferably aircraft cable wire) is wrapped about the base or platform of the robot as can be seen in FIG. 23. By activating one or the other of the first set of clutches, the cable rotates the robot in either the clockwise or counterclockwise direction.

(93) The second set of clutches receives torque from the first set of clutches and also transfers torque to the third set of clutches. As well, the second set of clutches moves the arm (coupled to the third set of clutches) up and down. This arm is dark colored in FIG. 22 and is between the first set of clutches and the third set of clutches. Again, by selecting one or the other of the clutches in the second set of clutches (illustrated in cut-away in FIG. 24 above the first set), the arm can be moved up or down.

(94) Referring again to FIG. 22, the third set of clutches at the topmost joint of the robot moves the end of the robot up or down. This third set receives torque from the second set of clutches and can be seen in FIG. 22 as having only two drive belts coupled to it. As with the other sets of clutches, this third set of clutches moves the end of the robot up or down depending on which clutch is active.

(95) Referring to FIG. 24, the cut-away view illustrates four clutch assemblies (in color) of the robot. These four comprise the clutches in the first and second sets of clutches. As can be seen, the two top clutch assemblies in cut-away have output drive belts coupled to the clutch assemblies at the top joint of the robotic arm. The robot in these Figures has the advantage of the Antagonistic-DASA configuration as well as the safety margin provided by the clutch assembly explained above. The Antagonistic-DASA configuration allows a single unidirectional drive motor to be used as a dual directional actuation unit. In the robot of FIGS. 22, 23, and 24, by simply judiciously activating and deactivating specific clutch assemblies, the robotic arm and its end can made to independently move up or down and the robot can be made to rotate about its vertical axis without having to change the direction of the drive motor. As well, all of the clutch assemblies provide the requisite safety margin in the event of an unintended collision between the arm and a human.

(96) A person understanding this invention may now conceive of alternative structures and embodiments or variations of the above all of which are intended to fall within the scope of the invention as defined in the claims that follow.