Kinematic and dynamic performance of prosthetic knee joint using six-bar mechanism

Abstract — Six-bar linkages have been used in some prosthetic knees in the past years, but only a few publications have been written on the special functions of the mechanism as used in transfemoral prosthesis. This paper investigates the advantages of the mechanism as used in the prosthetic knee from the kinematic and dynamic points of view. Computer simulation and an experimental method were used in the investigation. The results show that the six-bar mechanism, as compared to the four-bar mechanism, can be designed to better achieve the expected trajectory of the ankle joint in swing phase. Moreover, a six-bar linkage can be designed to have more instant inactive joints than a four-bar linkage, hence making the prosthetic knee more stable in the standing phase. In the dynamic analysis, the location of the moment controller was determined for minimum value of the control moment. A testing prosthetic knee mechanism with optimum designed parameters was manufactured for experiments in the laboratory. The experimental results have verified the advantage revealed in the analyses.

Abbreviations: CCD = charge-coupled device, IIJ = instant inactive joint.
This material was based on work supported by the Nature Science Fund No. 39770214 and No. 30170242 of China.
Address all correspondence and requests for reprints to Dewen Jin, Rehabilitation Engineering Center, Department of Precision Instruments, Tsinghua University, Beijing 100084, PR China; tel: 8610-62794189; email: jdw-om@tsinghua.edu.cn.

Four-bar mechanisms have been widely used in the prosthetic knee for many years and are a subject of investigation by Zarrugh, Radcliffe, Hobson, and other scientists and researchers [1-4]. Six-bar mechanisms have been successfully used in some knee joints, such as Total Knee and 3R60 Knee produced by the Otto Bock Company; a few publications on kinematic and dynamic performance of the six-bar knee mechanism have been reported [5,6]. The general constitution of multiple-bar linkage for the prosthetic knee was outlined by Van de Veen [5], but no further investigations have been reported. Patil and Chakraborty designed a particular six-bar knee-ankle mechanism to provide coordinate motion between knee and ankle joint during walking and squatting [6].

Compared with four-bar mechanisms, six-bar mechanisms have much more design variables. Therefore, with appropriate design, six-bar mechanisms can provide advantages that are more functional. The basic concerns with kinematic and dynamic analyses of a prosthetic knee include the gait pattern (especially the trajectory of ankle joint in swing phase, which provides enough foot ground clearance), angular displacement of the shank, and stability in the standing phase. Moreover, with the intelligent knee developed in the last several years, the desire has been to adapt the prosthesis to walking speed and terrain [7,8]. Therefore, it is necessary to adjust and control the knee moments according to the walking pattern. The values of control moments are the main considerations for developing the moment controller to make it suitable to the prosthetic knee.

In this paper, the kinematic and dynamic performance of the six-bar mechanism used in the prosthetic knee is investigated by computer simulation and some experiments. First, the constitutions of six-bar linkages with total revolute joints are stated. Second, the optimum design procedure is adopted for kinematic design to realize the expected trajectory (spatio-temporal curve) of the ankle joint. Moreover, because more Instant Inactive Joints (IIJs) can exist in six-bar mechanisms than can exist in four-bar mechanisms [9], the stability in the standing phase can be ensured even under some disturbance. For adaptability of the prosthesis to walking speed and terrain, the control moments in the swing phase were investigated in dynamic analysis. The results show that the control moment of the knee joint can be reduced considerably by an appropriate axis being chosen where the moment controller is located. Based on the results of the investigations, a testing prosthetic knee mechanism was manufactured for experimental use in the laboratory. Both analytical and experimental results given in this paper indicate that the advantages of the six-bar linkage can be achieved.

Fundamental types of six-bar mechanisms are the Watt type and Stephenson type as shown in Figure 1. Based on these two types, the knee joint has four configurations (see Figure 2(a) to (c)). The design parameters of these configurations are the same. The particular objective is to constitute the six-bar knee mechanism so that the shank is fixed to link 5 or 6 while the thigh is fixed to link 1. Otherwise, for example, if the shank is connected to link 3, then the function of the six-bar knee mechanism will be the same as that of four-bar mechanisms.

Figure 2.
Configurations (a) 1, (b) 2, (c) 3, and (d) 4 of six-bar mechanism for prosthetic knee.

The kinematic design aims to achieve the expected trajectory of the ankle joint and the locus of the geometric center of the knee mechanism and to ensure the stability in the extended position of the knee. Meanwhile, the dimensions of links should be within an acceptable range. The geometric center of the knee mechanism can be calculated by the equations

where

are the coordinates of the geometric center of the knee mechanism and

are the coordinates of the seven joints of the mechanism.

To meet the requirements just mentioned, we adopted the optimum procedure. The optimization is based on the expected relative motion of thigh and shank. As an example, taking the configurations shown in Figure 2(a) with the shank and link 5 connected (Figure 3), the optimization problem is expressed in the subsequent paragraphs.

where n is the number of selected points in a gait cycle, n = 25;

are the calculated coordinates of the trajectory of the ankle joint during the optimum process;

are the coordinates of the expected trajectory of the ankle joint; xKi , yKi are the calculated coordinates of the trajectory of the geometrical center of the knee joint during the optimum process;

are the coordinates of the expected trajectory of the knee joint; and C1,C2 are the weight factors and C1 + C2 = 0.9 + 0.1 = 1. C1 is much larger than C2 here, because emphasis is put on the locus of the ankle joint.

How to choose the expected trajectory is a problem needed to make further studies. What we used here is based on the gait analysis of the sound side of a transfemoral prosthesis user while walking at a normal speed (1.2 m/s), because we hope to increase the level of symmetry of gait parameters.

By defining a frame x O y fixed on the thigh, shown in Figure 3, the design parameters can be expressed as a vector X such that

The variables in the vector are as indicated in Figure 3. There are, in total, 16 elements, including 14 dimensions of links and two angular positions of the thigh and shank

and

respectively. The coordinates of the points A, B, C, D, E, F, G, I, J, and P in the frame are expressed as functions of the design parameters in the following equations:

where l is the distance between points R and S and x is the angle between the two lines S-R and T-R. Equation (16) is used to calculate the coordinates of an arbitrary point R(x ,y) based on coordinates of the other two known points S(x1, y1) and T(x2, y2).

where li is the same as defined in Equation (2) and limin and limax are the dimension limitation to the length of each bar.

where the limited values of the design variables lmin, lmax, Xmin, and Ymin were given based on the required size of the mechanism.

After the optimization method of Complex Penalty Function is applied, the design parameters are obtained as

Then the six-bar knee mechanism was designed, and the trajectory generated by the mechanism can be obtained by the kinematic analysis being appied.

The comparison of the generated trajectory of the ankle joint with expected ones is shown in Figure 4. The mean square errors for ankle and knee are Errankle = 1.96% and Errknee = 11.43%, respectively. The comparison of the trajectory of the ankle joint in swing phase of the six-bar linkage knee with that of a four-bar knee mechanism is also made and given in Figure 5. The dimensions of the four-bar linkage were designed with the use of the same procedure as that used for the six-bar linkage knee. The mean square error of ankle joint trajectories of the four-bar mechanism is 6.71 percent, while that of the six-bar mechanism is 1.96 percent.

Figure 5.
Comparison of ankle joint trajectory in swing phase between different mechanisms.

Moreover, when the user is walking on different terrain, such as on a slope or in different speeds, the six-bar knee mechanism possesses advantages from the kinematic point of view. The comparison is shown in Figure 6(a) to (c), and the mean square errors are listed in the Table.

Figure 6.
Trajectory of ankle joint in swing phase in different walking pattern: (a) fast walking, (b) up hill, and (c) down hill.

Table.

Mean square errors of ankle joint trajectories in different patterns (%).
Walking Pattern	Four-Bar	Six-Bar
Fast	5.16	2.27
Slow	6.71	1.96
Up Slope	9.87	3.97
Down Slope	11.9	4.61

In the multibar kinematic chain, if two links connected by a revolute joint have the same angular velocity in this position (or at this instant time), which means that no relative motion exits between these two links, the joint is referred to as IIJ. For example, if the four-bar kinematic chain, shown in Figure 7(a), is in such a position that links 3 and 4 are collinear and links 1 and 2 have the same angular velocity, then the revolute joint A is an IIJ

Figure 7.
Instant inactive joint of four-bar mechanism (P is instant velocity center; subscripts represent the number of bars): (a) four-bar kinematic chain and (b) four-bar mechanism with link 1 fixed.

in this position. The IIJ must exist for the mechanism to be self-locking. In Figure 7(b), if link 1 (or 2) is fixed, link 2 (or 1) cannot drive the mechanism no matter how large the driving moment is. Obviously, the more IIJ exists, the more stable the mechanism is. In the four-bar kinematic chain, only one IIJ can exist. However, in the six-bar kinematic chain, as many as four IIJs exits, depending on the design. For example, when links 2 and 3 of the six-bar kinematic chain are collinear as shown in Figure 8, the P14, P16, P45, and P56 joints will be IIJs. Therefore, when link 1 is fixed, the mechanism will be stable despite any disturbance applied on links 4, 5, or 6. In the optimum design stated in the last section, the constraint (equation 17) means links EF and FG (Figure 3) are collinear in the extended position of the knee. In this case, the A, B, C, and D joints are IIJs. Therefore, in addition to the functional advantage in kinematic design mentioned in the last section, the six-bar linkage for the prosthetic knee has another advantage based on stability.

Figure 8.
Instant inactive joint of six-bar linkage (P indicates the instant velocity center; subscripts represent the number of bars).

Kinematic design is based on the expected relative motion between thigh and shank. To realize the expected absolute motion of the shank and corresponding ankle joint trajectory requires not only kinematics but also dynamic analysis and control. One could obtain the expression for the control moment by using the inverse dynamic procedure. The control moment is usually applied on the knee joint. Essentially, the small control moment will be easier to be realized than large ones.

To determine the suitable axis where the control moment is to be applied, we performed the dynamic analysis for the mechanism to derive the control moment. Figure 9 depicts the free body diagrams for dynamic force analysis. To demonstrate the dynamic process, taking axis A (Figure 3) as an example (i.e., the control moment is supposed to be applied on joint A) and considering the balance of the moments and forces for each body (Figure 9), the procedure to determine the moment can be obtained by

Figure 9.
Free body diagram of six-bar mechanism when torque applied on joint A.

where

, denote the distances between G and T in x,y directions, respectively;

, denote the distances between G and S in x,y directions, respectively; ms, Js are the mass and the moment of inertia of the shank and foot respectively; asx ,asy and..q s represent the acceleration of the mass center S in x,y directions and angular acceleration of the leg, respectively. They were obtained from the expected movement of the shank.

is the perpendicular distance from point G to line EF and similarly to other points and lines, and

and MA are indicated in Figure 9. Finally, the control moment MA is derived from the equation

By applying the same procedure, the control moments MB, MG, and MF can be determined, too. The calculated results of MA and MB, MG, and MF in normal walking speed are plotted in Figures 10 and 11, respectively. The values in the vertical axis are the moments divided by the weight of the body W. One can observe that the values of MA are extremely larger than those of MB, MG , and MF, whereas MF is the smallest. Therefore, selecting axis F as the place where the control moment should be applied would be most appropriate.

Considering that the control moments depend on walking speed and terrain, such as going up or down hill, analyzing them in different walking situations is necessary. The moments required at axis F in the swing phase in different walking situations are shown in Figure 12.

In a practical prosthetic knee, usually the total control moment consists of two parts: the positive extension moment, acting to extend knee provided by an extension assist spring, and the negative resistance moment, produced by a damper. The control moment shown in Figure 12 is the total knee moment. To derive the moment of the damper that is controllable, one must minus the spring toque from the total moment. The resulting moment provided by the controllable damper is shown in Figure 13.

Based on the analysis just mentioned, a prosthetic knee mechanism was manufactured for experimental use (Figure 14). It consists of a six-bar linkage, a friction damping moment generator whose moment can be controlled by computer through a step motor (see Dewen Jin et al. for details [8]).

In the experiments, the prosthetic thigh was strapped to the thigh of a nonamputee subject to make the thigh move as close as possible to normal gait. It was dragged along in normal walking speed. The movement of the shank and the trajectory of ankle joint were recorded with a CCD (charge-coupled device) based human motion detecting and analysis system, which was developed at Tsinghua University [10]. The block diagram is shown in Figure 15.

The effectiveness of the analyses and design presented in this paper was evaluated with the use of the mean square error of the trajectory of the ankle joint Era (mean square errors = up slope 5.97%, down slope 5.73%, level [fast] 5.67%, and level [slowly] 4.06%) and the angular displacement of shank Ers (up slope 5.40%, down slope 4.17%, level [fast] 4.39%, level [slowly] 2.46%), by

where x,y and

are experimental values,

and

are expected values, and

(1) are reference values.

The six-bar prosthetic knee mechanism has been investigated from kinematic and dynamic points of view in this paper. The performance of the knee mechanism is shown in the following aspects:

· The values of control moments in swing phase were found to vary in a very large range when taking different axes as the place where the controller is located. The dynamic analysis is important for determining the most suitable place for knee moment control.

The knee mechanism was developed experimentally to show the feasibility of the procedures used in the investigation. To develop it for clinical application, further development (such as the reduction of the size and weight of the step motor, development of a smaller electric circuit with battery, the use of light and high strength material, etc.), should be undertaken. Furthermore, determining what is the most appropriate expected trajectory that can increase the level of symmetry of gait pattern remains an interesting and systematic subject. It has been found that the gait parameters of the sound side of a transfemoral prosthesis user are also affected by the prosthesis side. The factors which affect the gait pattern include not only the kinematic and dynamic performance of the prosthetic knee but also the construction of the ankle joint, functions of prosthetic foot, the quality of alignment, and the physical and psychological conditions of the user. Therefore, to obtain the most appropriate expected trajectory, further studies and clinic tests are needed. This paper focuses on the kinematic and dynamic design of the six-bar prosthetic knee mechanism. Certainly, the proper design of the prosthetic knee mechanism will be helpful to improve the level of symmetry of gait pattern.

7. Nakagawa A. Intelligent knee mechanism and the possibility to apply the principle to other joint. Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology; 1998 Oct. 29-Nov 1; Hong Kong, China. p. 2282-87.

8. Dewen Jin, Ruihong Zhang, Jichuan Zhang, Rencheng Wang, William A. Gruver. An intelligent above knee prosthesis with EMG based terrain identification. Proceedings of 2000 IEEE International Conference on System, Man and Cybernetics; 2000; Nashville, Tennessee. p. 1859-64.