Proceedings of the 2018 Design of Medical Devices Conference DMD2018 April 9-12, 2018, Minneapolis, MN, USA DMD2018-6822 THE APPLICATION OF SERIES ELASTIC ACTUATORS IN THE HYDRAULIC ANKLE-FOOT ORTHOSIS Jeong Yong Kim University of Minnesota, Twin Cities Minneapolis, MN, USA William Durfee University of Minnesota, Twin Cities Minneapolis, MN, USA ABSTRACT Advances in ankle-foot orthosis (AFO) technology have been trending toward more powerful lightweight devices. A hydraulic series elastic actuator (HSEA) was explored to design a lightweight powered AFO that meets the high peak power dem of ankle gait. With its excellent power density its ability to separate the power supply from the actuator using a hose, hydraulic power was used, combined with an SEA that takes advantage of the high-peak low-average power profile of ankle gait to store energy release it during the push-off stage of gait. The parameters required for the SEA were determined validated using simulation. A gait pattern that would require 235W of motor power was able to be tracked using a motor rated at 95W. The actuator weight of the hydraulic ankle-foot orthosis (HAFO) at the ankle was 0.35, which is 43% of an equivalent electromechanical system. A novel design of an HSEA with a clutch capability is proposed for future HAFO applications. INTRODUCTION Orthotic technology has been trending towards designing more powerful lightweight devices. Within this trend is the ankle-foot orthosis (AFO), which is a device worn at the ankle to assist users in gait impairments such as foot drop. In the past, AFOs were passive devices intended to restrain the movement of the ankle or passively provide stiffness for the user to bear weight onto. With advances in actuation technology, researches have designed powered AFOs that provide positive power to the ankle since strived to develop AFOs that provide more power with lighter systems. [1] AFOs need not provide the entirety of the power demed by gait depending on their purpose. AFOs intended for rehabilitation require the patient to use their muscle the AFO provides a portion of the power during gait. Powered AFOs used in a confined setting such as a hospital may not require the entire device to be lightweight as the power supply may be tethered to the ground, eliminating the need for the user to bear the weight. Despite the various purposes means an AFO can be designed for, there will be a continuous dem for a lighter more powerful device. AFOs will exceed the role of supporting a weakened joint to providing superhuman strength speed for workers in construction sites or firefighters in burning buildings. Furthermore, the actuation technology is applicable to other joints, opening up possibilities for whole-body exoskeletons. In the context of this trend, the objective of this study was to use a series elastic actuator (SEA) to lower the power wattage of the motor reduce the weight of the hydraulic ankle-foot orthosis (HAFO). The parameters for the SEA were determined based on a past study verified through simulation. Then the weight of different hydraulic electromechanical systems using direct drive SEAs were evaluated for the feasibility of a lightweight powered AFO. A hydraulic series elastic actuator (HSEA) was designed to be used in a future version of the HAFO. HYDRAULIC POWER Hydraulic technology has properties that make it advantageous for applications in wearable devices. It has potential to lower the weight of an AFO while generating high peak power that the ankle requires. Research has shown 100W small-scale hydraulic systems are more efficient lightweight when compared to electromechanical systems when operated above 500 psi [2]. Additionally, hydraulic cylinders can be separated from its power source through hoses to minimize the weight at the ankle. The effect of weight at the ankle shows increased rates of oxygen consumption which goes against the purpose of an AFO [3]. The advantages of fluid power in designing a lightweight AFO have been explored in a previous study by Neubauer Durfee [4]. Their HAFO was able to reduce the weight of the device at the ankle to less than 1.2 kg the overall weight to 3.5 kg. The HAFO achieved this weight with an untethered 1 Copyright 2018 ASME
power source a compact size that fits within a pant leg. It was able to output a portion of the power required by gait of a 56 kg male walking at a pace of 1 cycle/sec. SEA An SEA has an elastic component attached in series to the actuator to reduce the stiffness with its environment. [5] The SEA goes against the traditional idea that a stiffer system is easier to control. Some advantages are that it is easier to control the force output is able to prevent damage to the environment of actuator by absorbing external impacts from the environment. Most importantly for the purpose of this study, the SEA is able to store energy in its elastic component release it to amplify the peak power output. The SEA has been used in orthoses for different reasons. [1] The earliest use was to vary the compliance of a passive ankle foot orthosis to optimize the elastic constant for each user by controlling the amount of deformation in the spring. [6] SEAs were later used as a method of storing releasing energy to better mimic the human tendon. [7] The capacity of an SEA to store energy is especially useful for the ankle as the kinetics kinematics of ankle during gait can be characterized as highpeak low-average power where the peak power is required for a short portion of the gait cycle. Conventional design of an actuator power supply requires components that satisfy the peak requirement. However, by using an SEA, research has shown the peak power can be amplified by capturing releasing energy in the series elastic component. [7] ELASTIC RATE AND MOTOR POWER The initial step toward analyzing the weight of a HAFO using SEA was determining the motor size capable of generating the peak power required by gait. The kinetic kinematic data of the ankle during gait of a 56 kg male walking at a pace of 1 cycle/per was used from the appendix data of Winter [8]. Holler derived an equation for the power required by a linear actuator given the angular velocity moment at the ankle for his electromechanical series elastic powered AFO. The rotational motion was transformed into linear motion via a lever applied to determine the elastic rate size of motor. For the HAFO, a pulley radius of 30 mm was used for transmission. The equation was derived from the following free-body diagram. Fig. 1 Free-body diagram of SEA The power required by the motor was derived as the following equation. All the parameters except the spring rate were determined by the rotational motion of the ankle transferred to linear motion. Therefore, the size of the motor was expressed as a function of elastic rate of the spring as shown in fig 2. Consider a case with an infinitely large elastic rate. Only the first term from the right side of the equation remain the power required by the motor simply equals the power dem of the ankle. When the elastic rate reaches zero, the system becomes overly compliant requires nearly infinite amount of power by the motor to generate enough power for gait. It can be expected that there is an optimal elastic rate between the extreme values that minimizes the power required by the motor. In between the extreme values for the elastic rate, the motor power decreased below the peak power dem of gait at 20,000 N/m reached a minimum motor power of 95 W with an optimized rate of 40,000 N/m. From this calculation, the motor power could be lowered as much as 40% of the peak power required by the ankle. Fig. 2 Power required by motor as a function of elastic rate SIMULATION A model of the HAFO was simulated using the Simscape Fluids package of Simulink. The purpose of simulation was to verify the spring rate motor power required from calculation its ability to amplify the peak power to meet the dems of gait. To isolate the effect of the series elastic component, two systems were compared: a direct drive system a series elastic system. The velocity-torque limitation for both systems was designed to be 95W based on the calculations in the previous section. A proportional-derivative (PD) controller was used to track the angular position of the joint while the moment on the ankle was applied as an external torque. Figure 3. illustrates the closed loop hydraulic circuit (1) 2 Copyright 2018 ASME
the linear motion from the cylinders transferred to rotational motion via pulley. Fig. 3 Closed loop hydraulic circuit of series elastic HAFO. The pulley axis is aligned to the ankle joint Fig. 5 Power required by gait power generated by DD SEA 95W systems The reference ankle position started from heel strike to the next heel strike. The kinetic kinematic data were of a 56 kg male walking at a pace of 1 cycle/per [8]. The peak power required occurred at 0.5 seconds when the ankle undergoes plantar flexion at a high angular velocity while simultaneously propelling the body weight forward. The moment as seen from the ankle was the result of the reaction torque from body weight the generated torque of the tendons the effect of moment was shown during push-off which was in the 0.4 to 0.8 region. It can be seen in fig 4. that the direct drive system was not able to track the position deviated from the desired position during push-off. The series elastic system deviated from the reference position in the 0.3 to 0.5 second region but was able to track the position well during push-off. WEIGHT EVALUATION The simulation results showed the high peak power of gait can be reached using an SEA coupled with an underpowered motor. The weight of a HAFO using the components in the simulation was analyzed to evaluate the amount of weight reduction that could be achieved. For a realistic solution, commercially available products from the same manufacturer line of products were compared. Five systems were compared. To demonstrate the weight advantage the HAFO, three hydraulic systems were compared to two electromechanical systems. The first system was the direct drive hydraulic system developed by Neubauer Durfee [4] that is not able to meet the entire dem of gait. The second system was a direct drive system but with a bigger motor designed to meet the peak power requirements of gait. The third hydraulic system was the series elastic hydraulic system using the derived component parameters. The electromechanical series elastic system was a powered AFO developed by Holler [7] that had similar performance to the first hydraulic system. The last system was an electromechanical system using direct drive able to generate the peak power of gait. Table 1. summarizes the component-wise weight of each system. For a direct comparison of hydraulic electromechanical systems, the weight of actuator power supply components of a powered AFO were compared excluding the frame miscellaneous parts of the assembly. The component weights of the five systems are visualized in bar graph form. In fig. 6, the component weights were divided according to which part of the body the weight is applied. All hydraulic systems were lighter when considering only the weight at the ankle because the power supply was located at the waist. Comparing direct drive systems with same performance, the hydraulic system was 0.35 kg when the electromechanical Fig. 4 Ankle position tracking comparison of direct drive series elastic systems using 95W motors The power profiles of the direct drive series elastic systems are shown in fig 5. The direct drive was not able to output the required power of 235W. The series elastic system was able to use the energy stored in the spring to reach the peak power of 235W. 3 Copyright 2018 ASME
Table 1. Component weight in kilograms of five systems Type Hydraulic EM Config. / Perform. Direct drive / Low Direct drive / Full Series elastic / Full Series elastic / Full Direct drive / Full Motor 0.14 (Maxon EC 45 f) 1.1 (Maxon EC 45) 0.47 (Maxon EC 60 f) 0.34 (Maxon RE 35) 1.1 (Maxon EC 45) Gearbox 0.11 0.26 0.26 0.11 0.26 Axial pump 0.27 (Takako 0.4 cc) 0.27 (Takako 0.4 cc) 0.27 (Takako 0.4 cc) - - Manifold 0.51 0.51 0.51 - - Cylinder(s) 0.35 0.35 0.35 - - Spring(s) - - 0.06 (LHP207L04S) 0.05 - Lead screw - - - 0.45 0.45 Total 1.38 2.49 1.92 0.95 1.81 system was 1.81 kg. The series elastic systems were 0.41 kg for hydraulic 0.95 for electromechanical systems. However, the overall weights were lighter for electromechanical systems as they do not require the added components of axial pump manifold. Among the three hydraulic systems, the series elastic system was heavier at the ankle due to the added components for series elasticity but lighter overall when comparing between systems with the same performance. HSEA DESIGN The simplest form of applying series elasticity to a hydraulic cylinder would be to attach a spring to the rod. The HSEA is a double acting hydraulic cylinder with a spring as its elastic component nested within the chamber. Figure 7. shows the cross section. The HSEA has three chambers the fluid ports connected to the outer chambers provide the fluid for plantar flexion dorsi flexion same as a traditional double acting cylinder. The additional middle port acts as a Fig. 6 Weight distribution of different systems between ankle waist clutch that enables or disables the series elasticity located in between the two pistons. When fluid is allowed to flow into the middle port the spring deforms stores or releases energy. When the port is blocked fluid flow into the port is not allowed, the spring is unable to deform due to the incompressibility of the fluid maintaining the deformation of the spring. In this state, the system becomes a direct drive system until the middle port is opened. Fig. 7 Cross section of HSEA in (a) retracted (b) extended positions Figure 8. illustrates the movements of the pistons when the ankle leads up to undergoes push-off. The direction of the pistons respect to each other is important. Energy is stored in the spring as force is applied in the negative direction while high pressure is supplied in the opposing direction. As the ankle pushes off the ground, switching from dorsi flexion to plantar flexion, the energy stored in the spring is released, in addition to the power supplied by the motor. At this moment, the cylinder is able to surpass the rated power of the motor to meet the dems of ankle gait power. CONCLUSION By applying SEA to the HAFO, simulation results showed the power requirements of gait were able to be met using a smaller motor compared to a direct drive AFO. A gait pattern with a peak power of 235W was able to be tracked using a motor of 95W. A component-wise weight evaluation was able to demonstrate the possibility of reducing the overall weight of the HAFO. The significance lies at the weight of the actuator at the ankle which was 43% of an equal performing electromechanical system. Additionally, a design concept for a HSEA with clutch capabilities was proposed for future prototypes. 4 Copyright 2018 ASME
Fig. 8 Cylinder piston movements during push-off. (a) spring is compressed (b) energy stored is released. The HSEA design is only in its concept level needs validation through benchtop testing. Acquiring commercially available hydraulic components remains a challenge to devices with small-scale hydraulics. The future of orthotic technology trends toward lighter device that can output more power, widening the applications of orthotic devices. REFERENCES [1] Herr, [2] [3] [4] [5] [6] [7] [8] H., 2009, Exoskeletons Orthoses: Classification, Design Challenges Future Directions, Journal of NeuroEngineering Rehabilitation., 6(1), pp. 21-30. Durfee, W., Xia, J. Hsiao-Wecksler, E., 2011, "Tiny Hydraulics for Powered Orthotics," Rehabilitation Robotics (ICORR), 2011 IEEE International Conference on 2011, pp. 1-6. Barnett, S., Bagley, A. Skinner, H., 1993, Ankle Weight Effect on Gait: Orthotic Implications, Orthopedics, 16(10), pp. 1127-1131. Neubauer, B., Durfee, W., 2016, Preliminary design engineering evaluation of a hydraulic ankle-foot orthosis, Journal of Medical Devices, Transactions of the ASME, 10(4), pp. 1-9. Pratt, G. A., Williamson, M. M., 1995, Series Elastic Actuators, Intelligent Robots Systems 95. Proceedings 1995 IEEE/RSJ International Conference on Intelligent Robots Systems, 0 1, pp. 399-406. Blaya, J., Herr, H., 2004, Adaptive control of a variable-impedance ankle-foot orthosis to assist drop-foot gait, Neural Systems Rehabilitation Engineering, IEEE Transactions on, 12(1), pp. 24-31. Holler, K. W., Ilg, R., Sugar, T. G., Herring, D., 2006, An efficient robotic tendon for gait assistance, Journal of Biomechanical Engineering, 128(5), pp. 788791. Winter, D., 1990, Biomechanics motor control of human movement (2nd ed.). New York: Wiley. 5 Copyright 2018 ASME