Target of physiological gait: Realization of speed adaptive control for a prosthetic knee during swing flexion

Abstract

BACKGROUND:

Prosthetic knee is the most important component of lower limb prosthesis. Speed adaptive for prosthetic knee during swing flexion is the key method to realize physiological gait.

OBJECTIVE:

This study aims to discuss the target of physiological gait, propose a speed adaptive control method during swing flexion and research the damping adjustment law of intelligent hydraulic prosthetic knee.

METHODS:

According to the physiological gait trials of healthy people, the control target during swing flexion is defined. A new prosthetic knee with fuzzy logical control during swing flexion is designed to realize the damping adjustment automatically. The function simulation and evaluation system of intelligent knee prosthesis is provided. Speed adaptive control test of the intelligent prosthetic knee in different velocities are researched.

RESULTS:

The maximum swing flexion of the knee angle is set between sixty degree and seventy degree as the target of physiological gait. Preliminary experimental results demonstrate that the prosthetic knee with fuzzy logical control is able to realize physiological gait under different speeds. The faster the walking, the bigger the valve closure percentage of the hydraulic prosthetic knee.

CONCLUSIONS:

The proposed fuzzy logical control strategy and intelligent hydraulic prosthetic knee are effective for the amputee to achieve physiological gait.

Keywords

Physiological gait speed adaptive prosthetic knee fuzzy logical control function simulation and evaluation

1. Introduction

According to the data calculation of the second national handicapped person sampling survey, the sum of disabled people in China is 82.96 million, including 24.12 million physical disabilities and 2.26 million amputees [1]. The number of lower limb amputee is about 1.58 million [2]. The residual limb regeneration cannot be reached in the current medical treatment level. The only way to restore the walking ability is to install lower limb prosthesis.

A complete prosthesis for above knee amputee is composed by: a socket, a knee prosthesis, an ankle-foot prosthesis and a link between the laters [3]. The part that has taken the biggest efforts in its development is the knee prosthesis. The knee performs a key role during the whole gait cycle, and this is basically due to the fact that the knee joint sets a leg made of two bodies, articulated between each other through it. This is the case in human body between thigh and shank, being this leg configuration from which it is understood a natural gait. Moreover, this last feature, leaving aside damping and impulse functions of feet, is what makes human gait the most efficient mode of moving human body from one place to another [4].

Innovation in prosthetic knee joint design, especially in recent years, has shown that functionality and thus the realization of physiological gait for persons who have undergone above knee amputations has increased. This can be attributed to the use of new technology and an improved understanding of the amputee biomechanics. The prosthetic knee is separated by the complexity of their control: mechanically passive, microprocessor controlled passive, and microprocessor controlled active [5]. Damping of mechanically passive control knee cannot be changed with different speeds and often characterized by abnormal gait, increasing energy expenditure of amputees to walk. Intelligent prosthesis are featured by having an on-board microprocessor which controls the actuator response. The actuator, in the passive ones may be a hydraulic or pneumatic cylinder controlled by opening or closing its valves through servomotors. For the active ones, flexion and extension of knee are driven by an external force such as the motor, micro pump and so on [6].

The focus of intelligent knee prosthesis is how to realize the physiological gait in swing phase under different speeds. It includes the research of physiological gait, prosthesis design and speed adaptive control in swing phase. BGA Lambrecht presented an autonomous hydraulically powered prosthetic knee. Operation modes include an active mode driven by a pump, and a passive mode controlled by a variable position valve. However, the hydraulic valve loop is extremely complex and heavy [7]. Seid designed a passive controller for hydraulic damper for swing phase of single axis knee; however, the controller resulted in a very large deviation of knee flexion angle from the normal one and hence the designed controller was reported to perform poorly in terms of ground clearance [8]. Hugh Herr provided a magnetorheological knee prosthesis which automatically adapted knee damping to the gait of the amputee using only local sensing of knee force, torque, and position [9]. Orhanli studied the finite state machine to control intelligent pneumatic knee in different phases through the comparison of physiological gait and prosthetic gait to evaluate the performance of the prosthesis [10].

The purpose of this study is to research the characteristic of physiological gait to define the target of prosthesis control in swing flexion. The similarity of hydraulic damping characteristics and knee joint damping torque is utilized to design a new structure of intelligent knee joint. Function simulation and evaluation system of intelligent knee prosthesis is developed. Speed adaptive control principle in swing flexion is studied according to the test device. Damping adjustment law of intelligent hydraulic prosthetic knee is revealed to realize physiological gait.

2. Methods

2.1 Physiological gait research

Figure 1 illustrates a single stride of the gait cycle. The stride begins with the heel strike of the leg. States 1 and 2 represent early stance flexion and extension just after heel strike (HS), respectively. State 3, or pre-swing, typically occurs at the end of stance, beginning just after the knee has fully extended and terminating when the foot has left the ground at toe-off (TO). States 4 and 5 represent periods of knee flexion and extension during the swing phase of walking, respectively.

Figure 1.

A normal gait cycle is shown schematically with state transitions represented. Adopted from [9].

The main function of prosthetic knee joint in stance phase, States 1–3, is to keep the stability. Speed adaptive is mainly in swing phase [11]. For State 4, the peak flexion angle is very important in amputee gait [12]. If this angle is too large, knee joint will not be extended before the next heel strike. To prevent tripping under this condition, amputees are forced to either walk slower or work harder to push the knee forward during swing extension. This will increase energy consumption and cause uncomfortable feeling. If, on the other hand, the knee does not reach a sufficient angle, there is an increased chance that the amputee will stub their toe on the ground when swinging it forward. Finally, a large difference in maximum flexion between an amputee’s healthy leg and the prosthesis is clearly visible and discordant. The primary goals of State 5 are to dampen swing sufficiently such that terminal impact is not too hard, while simultaneously not taking too much energy out of the swing such that the subject has to use excessive energy to extend the leg in time for heel strike. For lower limb prosthesis, the damping of swing extension is often held constant at a fairly high value [13]. Therefore, the damping adjustment of this research is focus on the speed adaptive in State 4. That means whether the maximum swing flexion can be achieved in different speeds.

Inman pointed out that maximum swing flexion angle typically does not exceed 70 degrees in normal walking [14]. Therefore, the goal of swing flexion adaptive control is to limit the maximum flexion angle of the knee joint. However, whether 70 degrees is also suitable for Asian people or energy-efficient still need more research. To define the target maximum swing flexion angle of intelligent knee prosthesis for Asian amputee. Healthy people are recruited for the study. Table 1 contains subject-specific information on personal data. Every subject is required to walk on the treadmill in three specific velocities, 0.5 m/s, 1 m/s and 1.3 m/s.

Table 1

Data on subjects

Subject	Sex	Age (years)	Height (cm)	Weight (kg)	Thigh length (cm)	Shank length (cm)	Foot size (cm)
1	Male	25	175	66	43	34	24
2	Male	24	182	75	46	37	25
3	Male	26	171	63	41	33	24
4	Male	23	165	57	39	32	23
5	Male	22	170	65	41	32	23
6	Female	23	163	53	40	32	22
7	Female	25	158	55	38	28	23
8	Female	22	160	56	39	29	22
9	Female	24	160	51	39	30	23
10	Female	24	159	49	38	28	23

The kinematic parameters are measured using an inertial 7 sensor system (GaitWatch, Jumho Electric, China). GaitWatch is consisted of gait acquisition instrument, gait analysis software and bandage. The important parameters of hip, knee and ankle are obtained with wireless sensors in real time. There is no need to use camera and professional laboratory, the test can be completed in common room and treadmill. When the wireless sensors are positioned on the subject, correction actions should be done to eliminate the error. The entire process is presented in the form of three-dimensional skeletal animation in software interface. The skeletal animation is just to show the movement vividly. It does not affect the measurement data. The position of the sensors, placement and program interface is as shown in Fig. 2.

Figure 2.

The position of the sensors, placement and program interface of GaitWatch.

Figure 3.

Maximum swing flexion angle of knee joint in different speeds.

Maximum swing flexion angle of knee joint recorded in different velocities of every person are shown in Fig. 3. The peak flexion angle (and as a consequence the maximum heel rise) during swing is 50 $\sim$ 70 degree for all the tested people in different speeds. Therefore, the target range of maximum swing flexion angle of knee joint is chosen between sixty and seventy degrees according to the test. It is safer to have slightly too much flexion than to risk applying too much resistive torque.

2.2 Technical background of the designed prosthetic knee joint

The microprocessor-controlled knee joint is designed and researched to further increase the range of prosthetic functions and applications available to prosthetic users. The microprocessor-controlled hydraulic knee joint functions through the use of simulated physiologic rule sets with auto-adaptive swing phase control.

2.2.1 Mechanical design

To realize the adjustment of the damping, the valve opening size is adjusted to change the hydraulic fluid velocity through drive motor of intelligent knee joint. The hydraulic damping force presents different properties with the change of the flow rate [15].The new knee joint is a homotaxial joint design, in which linear hydraulics (Fig. 4a) generate the necessary joint resistance for the individual gait phases and functions. No support for movements by actuators is needed. The hydraulics have two separately adjustable needle valves, whose opening diameters are set by two servomotors that control the internal flow resistance. To convert the immersion and extraction resistance to torque, the linear hydraulic system is connected to the upper unit of the joint by a geometric lever arm (Fig. 4b).

Figure 4.

(a) Integrated linear hydraulics, (b) Prosthetic knee joint component.

Figure 5.

Ankle pylon designed for the knee joint. The pressure sensors are FSR402 (Force Sensing Resistor 402) bought from Interlink Electronics.

Figure 6.

Pressure signal change process in different phases.

2.2.2 Sensor system

To implement various functions, it is necessary to use a number of sensors (knee angle sensor, axial load sensor, ankle pressure sensor, tri-axial accelerometer), which provide the information needed to control joint resistance [16]. In addition, to joint positions and movements, the parameters they detect also include acting forces. Some of these parameters are converted into other physical quantities. Most of the sensors are integrated directly into the knee joint. In addition, loading sensors and ankle pressure sensor are built into the tube adapter that connects the knee joint with the prosthetic foot. Two ankle pressure sensors are used to distinguish swing phase and stance phase. A special ankle is designed to place the load sensor and ankle sensor, as is shown in Fig. 5. The placement makes it that the change of artificial foot has no effect on the gait detection. The pressure signal change process in different phases is shown in Fig. 6. The initial state is (0, 0); when the heel contacts the ground, the back pressure sensor is squeezed and produces the signal. The state changes to be (1, 0). In the foot contact phase, back and forward pressure sensors are forced and the state is (1, 1). Before toe off, the signal changes to be (0, 1). When the signal is (0, 0), the swing flexion and extension are identified through the direction of the knee joint angular velocity.

Figure 7.

Contact time (from heel strike to toe off) vs. walking speed for normal subjects. Adopted from [13].

2.3 Speed adaptive control during swing flexion

Adaptation for swing extension is based on the concept of using enough deceleration to decrease terminal impact while ensuring that the knee always reaches full extension before heel strike [17]. For this article, swing extension valve is set in a fixed position. The adjustment of the valve is adaptive in swing flexion. To realize the speed adaptive control. It includes two aspects: speed identification and the choice of control law. A possibility for approximating walking speed is to look at the contact time for the stance phase preceding a given swing phase. Biological data shows a strong inverse correlation between contact time (time from heel strike to toe off) and walking speed, as is shown in Fig. 7 [18].

The auto-adaptation for swing flexion is designed to limit the maximum flexion angle for swing. The prosthetic knee joint and wearer is a nonlinear system [19]. Fuzzy logic control is easy to get good control in the nonlinear system with simple fuzzy inference [20]. Human walking is an unstable, strong coupling and nonlinear system, which is suitable for fuzzy rules to control. The idea of control algorithm is to compare the differential of contact time for the stance phase in the sequential gait cycle with error threshold to control the valve position. The control block diagram of swing flexion is shown in Fig. 8.

Figure 8.

Control block diagram of swing flexion.

When the error absolute value is less than the set value, keep the gait speed:

If $|\text{T}_{\text{n}}-\text{T}_{\text{n}-1}|<\text{E}_{\text{t}}$ then $\text{K}_{\text{n}}=\text{K}_{\text{n}-1}$ ;

When the error is greater than the set value, gait velocity decrease

If $|\text{T}_{\text{n}}-\text{T}_{\text{n}-1}|>\text{E}_{\text{t}}$ then $\text{K}_{\text{n}}=\text{K}_{\text{n}-1}-\text{AE}_{\text{k}}$ ;

When the error is smaller than the set value, gait velocity increase

If $|\text{T}_{\text{n}}-\text{T}_{\text{n}-1}|<\text{E}_{\text{t}}$ then $\text{K}_{\text{n}}=\text{K}_{\text{n}-1}+\text{AE}_{\text{k}}$ ;

$\text{K}_{\text{n}}$ : valve position calculated in the nth gait cycle; $\text{K}_{\text{n}-1}$ : valve position calculated in the ( $\text{n}-1$ )th gait cycle; $\text{T}_{\text{n}}-\text{T}_{\text{n}-1}$ : differential of contact time for the stance phase in the sequential gait cycle; A: gain coefficient; $\text{E}_{\text{t}}$ : time error threshold; $\text{E}_{\text{k}}$ : the minimum adjustment value of valve position; a: 5 degree.

The gain coefficient A is adjusted through the fuzzy logic control. When the input error is larger, the bigger gain coefficient is used to increase the rate of convergence. When the input error is smaller, the lesser gain coefficient is used to ensure the stability of the control.

2.4 Function simulation and evaluation system of the prosthetic knee

Test trials of intelligent lower limb prosthesis may be not safe and inconvenient to attempt with patients [21]. Robotic test can facilitate the development of new concepts, designs and control systems for prosthetic limbs [22]. In order to evaluate the gait performance of the intelligent hydraulic knee, the function simulation prototype of prosthetic knee is designed. It has two main functions: (1) Simulation of velocity changes and the real-time detection of knee angle; (2) Simulate the hip drive of the prosthetic knee to evaluate the swing performance. The composition and working principle of the device is shown in Fig. 9.

Figure 9.

Schematic diagram of function simulation and test system.

The whole control circuit of the function simulation and evaluation prototype includes three control modules:

Cylinder control module of the leg height adjustment. Height adjustment cylinder is controlled by a logic circuit board. Self-locking cylinder which is controlled by integrated circuit 4013CMOS is also used to prevent the prosthetic leg to fall into the ground and improve the location accuracy.

Starting module of step motor driver and brushless electric machine driver. The circuit board module adopted the AD654 integrated circuit to produce the drive signal of step motor.

Automatic control module. This module uses the microprocessor AT89S52 to realize signal detection and automatic control. Simulators communicated with desktop computer, microcomputer control and detection are achieved directly.

The designed system uses synchronous belt to simulate level walking of the prosthesis and lift air cylinder to simulate the human body’s gravity shift. Intelligent lower limb prosthesis is connected with the simulator by a special joint to realize the function simulation.

Figure 10.

Adaptive control of swing-flexion damping at 0.5 m/s.

Figure 11.

Adaptive control of swing-flexion damping at 1 m/s.

Figure 12.

Adaptive control of swing-flexion damping at 1.3 m/s.

3. Speed adaptive tests of the prosthetic knee during swing flexion

In Figs 10–12, both the maximum swing flexion angle and the electronic knee damping values (represented by valve closure percentage) are plotted against the number of walking steps taken from walking speeds for 0.5 m/s, 1 m/s, 1.3 m/s, respectively. During the first 7 walking steps, the maximum swing flexion angle is not in the range of sixty to seventy degrees. Consequently, the user-adaptive knee controller increases flexion damping until the maximum flexion angle falls in the range of biological threshold.

When the damper valve closure is 0%, maximum swing flexion angle of knee prosthesis are 86 ${}^{\circ}$ , 91 ${}^{\circ}$ and 97 ${}^{\circ}$ in 0.5 m/s, 1 m/s and 1.3 m/s, respectively. It means that when the flexion damping valve is fully open, the maximum swing phase knee flexion angle increase with the speed enlargement. When the damper valve closure changes, the maximum swing-phase knee flexion angle can be adjusted to 60 ${}^{\circ}$ to 70 ${}^{\circ}$ under different velocities. The valve closure percentage are 25%, 40% and 70% to realize the physiological gait in 0.5 m/s, 1 m/s and 1.3 m/s, respectively. It means that the valve closure percentage is changing to adjust to physiological gait under different speeds. The greater the speed, the bigger the closure percentage, that is, the smaller the valve opening.

4. Discussion

Target of physiological gait is more than one. Gait symmetry, maximum swing phase knee flexion angle and the energy consumption are also important indicators. Maximum swing phase knee flexion angle is chosen to be the control object. The fuzzy logic control has achieved good result for this target. However, the other indicators such as its application in the gait symmetry effect remains to be further studied.

Although function simulation and test device can improve the objectivity of the performance evaluation of the knee prosthesis, the study of traffic is relatively single. The level walking is simulated using synchronous belt. However, the performance of the knee prosthesis in complex road environment such as uphill and downhill, up and down stairs and overcome obstacle remains to be studied.

Knee joint acts as damper in the process of normal walking most of the time. However, it also needs active force under the condition such as up the slope and stairs. Although prosthesis has good damping performance, it still cannot provide active torque. The combination of active source and passive damper is an important research direction. In a follow-up study, motor or micro pump driven is connected with the current damper to further improve the imitation of nature.

5. Conclusions

Based on the physiological gait trials, the maximum swing flexion of knee is chosen as the control target. The new intelligent hydraulic knee prosthesis has been designed. A novel and low cost ankle pylon is developed to detect stance and swing phase. To simulate and evaluate the designed prosthetic knee, the robotic test prototype is proposed. It can guarantee safety and increase repeatability compared with human trials. The fuzzy logical control is suitable for adaptive control of intelligent prosthetic knee during swing flexion from the tests. In addition, more advanced control techniques will also be studied and implemented on the knee prosthesis.

Footnotes

Acknowledgments

The work reported in this paper is supported by National Natural Science Foundation of China, number: 61473193 and Shanghai Engineering Research Center of Assistive Devices, number: 15DZ2251700.

Conflict of interest

None to report.

References

Zhao

. China disabled persons’ federation release the latest disabled population data [J]. Disability in China, 2012, (4).

Yang

Liu

Geng

. Research advance on key technology of intelligent lower limb prosthesis [J]. Journal of Hebei University of Technology, 2013(1): 76-80.

Lambrecht

BGA

. Design of a hybrid passive-active prosthesis for above-knee amputees [J]. Dissertations & Theses – Gradworks, 2008.

Khan

Usman

. Effects of toe-out and toe-in gait with varying walking speeds on knee joint mechanics and lower limb energetics [J]. Gait & Posture, 2017, 53: 185.

Rosa

. Controller Design and Implementation for a Powered Prosthetic Knee [J]. Dissertations & Theses – Gradworks, 2012.

. The present situation and development trend of the artificial knee prosthesis [J]. Chinese Journal of Rehabilitation Theory and Practice, 2016, 22(9): 1080-1085.

Lambrecht

BGA

Kazerooni

. Design of a semi-active knee prosthesis[C]// IEEE International Conference on Robotics and Automation. IEEE, 2009: 639-645.

Seid

Sujatha

Chandramohan

. Design of controller for single axis knee using hydraulic damper[C]// Africon. IEEE, 2015: 1-5.

Herr

Wilkenfeld

. User adaptive control of a magnetorheological prosthetic knee [J]. Industrial Robot, 2003, 30(1): 42-55.

10.

Orhanli

Yilmaz

. Application and comparison of finite state control of above-knee prosthesis with pneumatic cylinder at different phases[C]// Signal Processing and Communications Applications Conference. IEEE, 2015.

11.

Sapin

Goujon

, et al. Functional gait analysis of trans-femoral amputees using two different single-axis prosthetic knees with hydraulic swing-phase control: Kinematic and kinetic comparison of two prosthetic knees [J]. Prosthetics & Orthotics International, 2008, 32(2): 201-218.

12.

Furse

Cleghorn

Andrysek

. Development of a low-technology prosthetic swing-phase mechanism [J]. Journal of Medical & Biological Engineering, 2011, 31(2): 145-150.

13.

Wilkenfeld, Ari Jacob. Biologically inspired auto adaptive control of a knee prosthesis [D]. 2000.

14.

Inman

Ralston

Todd

. Human Walking [M]. 1981.

15.

Ling

. Design of electronically controlled hydraulic damper for prosthetic knee [J]. Journal of Clinical Rehabilitative Tissue Engineering Research, 2009, 13(39): 7635-7638.

16.

Escalona

Torrealba

. Design of linear hydraulic actuator of variable response for intelligent knee prosthesis[C]// V Congreso Venezolano De Bioingeniería Bioven. 2015.

17.

Mooney

. Variable damping controller for a prosthetic knee during swing extension [D]. Massachusetts Institute of Technology, 2012.

18.

Wilkenfeld

Herr

. An Auto-Adaptive External Knee Prosthesis [J]. Artificial Intelligence Laboratory, 2000.

19.

Zhang

Wang

, et al. Design of a novel bionic prosthetic knee joint [J]. Assembly Automation, 2016, 36(4): 398-404.

20.

Chu

Feng

. Nonlinear system control with fuzzy logic design[C]// Fuzzy Systems, 1994. IEEE World Congress on Computational Intelligence. Proceedings of the Third IEEE Conference on. IEEE, 1994: 1592-1597 vol.3.

21.

Sinha

VDH

Arokiasamy

. Factors affecting quality of life in lower limb amputees [J]. Prosthetics & Orthotics International, 2011, 35(1): 90-96.

22.

Dingwell

Davis

Frazier

. Use of an instrumented treadmill for real-time gait symmetry evaluation and feedback in normal and trans-tibial amputee subjects [J]. Prosthetics & Orthotics International, 1996, 20(2): 101.