Simulation and Experimental Analysis of a Portable Powered Ankle-Foot Orthosis Control

A portable powered ankle-foot orthosis (PPAFO) was previously developed using off-the-shelf pneumatic components to explore new opportunities for fluid power in human assist devices. The untethered pneumatically powered ankle-foot orthosis provides both motion control and torque assistance at the ankle via a binary, event-based control scheme that uses solenoid valves. While stable, the binary actuation of the solenoid valves that results from this approach limits the overall performance of the system. This paper addresses the limitations of the current system using a modeling approach for both hardware and control design. Hardware and control configurations were first evaluated using simulations of the modeled PPAFO and shank-foot system during a simplified functional gait task: assistive propulsive torque during stance. These simulations demonstrated that the introduction of a proportional valve and new control architecture resulted in PPAFO performance improvements during the task. These results were then confirmed experimentally with the PPAFO attached to a physical model of a shank and foot. INTRODUCTION Ankle foot orthoses (AFOs) are orthotic devices used to correct lower limb gait deficiencies. Sizable populations exist in the United States alone that present with symptoms that could be corrected with AFOs: stroke (8M), spinal cord injuries (1.3M), multiple sclerosis (1M), cerebral palsy (412K), and polio (272K) [1, 2]. The ideal AFO should accommodate the many aspects of gait affected by injury or pathology, while being compact and light weight to minimize the energetic impact to the wearer [3]. Prescription AFOs are generally passive due to the size constrains, since the necessity of power source and actuator for active AFOs always results in increased size. Passive AFOs provide assistance by preventing unwanted foot motion with direct physical resistance [4, 5]. Although the motion control provided by passive AFOs can improve functionality, passive devices have limitations that affect performance. For example, the static nature of passive AFOs can impede gait by restricting movements that the patient is normally capable of attaining. Additionally, these devices are unable to adapt to changing environmental conditions or provide supplemental torque [6]. To address these limitations, we have developed a lightweight, portable powered ankle-foot orthosis (PPAFO) using mostly off-the-shelf components [7, 8]. Pneumatic power was selected because of its high power to weight ratio and low power consumption during static forcing. Currently, the torque assistance is controlled with two solenoid valves using binary (all on or all off), event-based control. Sensor measurements are used to identify events during gait and trigger corresponding valve configurations. Solenoid valves were initially selected for their size, simplicity, low cost, and low electrical power consumption. While this approach does provide supplemental torque assistance, there are limitations with the current system. This paper addresses two of these limitations: (1) the binary control of the solenoid valves does not allow the PPAFO to provide intermediate levels of torque Proceedings of the ASME 2011 Dynamic Systems and Control Conference DSCC2011 October 31 November 2, 2011, Arlington, VA, USA 1 Copyright © 2011 by ASME DSCC2011-6119 Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/19/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use assistance, and (2) the current control scheme may result in excessive actuation and high pneumatic power consumption, limiting the duration of use. This paper will address the performance and efficiency limitations related to the solenoid valves through a model-based system analysis of a new proportional valve configuration and improved control design. Hardware and control configurations were evaluated using simulations of the modeled PPAFO and shank-foot system during an idealized functional gait tasks: provide assistive propulsive torque during stance. The results from the simulations were then confirmed experimentally with the PPAFO and a physical model of a shank and foot. The contributions from this paper are twofold. First, this work provides an illustrative example of how to effectively utilize a well-identified system model to aid hardware design as well as develop and test different control strategies for a robotic assist device. Second, the model of the pneumatic system developed here can be used by other researchers working with pneumatic systems. The remainder of this paper will describe the current PPAFO hardware, present models of the PPAFO system and simplified leg, implement the hardware and control schemes both in simulation and experimentally, and compare the performance of the PPAFO with solenoid and proportional valve configurations during an idealized gait task. FIGURE 1. (A) GAIT IS CYCLIC AND SUBDIVIDED INTO MULTIPLE PHASES DEFINED BY FUNCTIONAL TASKS [9]. THE PORTABLE POWERED ANKLE-FOOT ORTHOSIS (PPAFO) ASSISTS GAIT BY CONTROLLING FOOT MOTION AND PROVIDING ASSISTIVE TORQUE FOR PROPULSION. (B) THE PPAFO IS POWERED BY A COMPRESSED CO2 BOTTLE (FAR RIGHT). METHODS PPAFO Hardware Walking consists of cyclic motion patterns that are divided into gait cycles beginning and ending at heel strike (Fig. 1A). Lower limb pathology or injury can disrupt the efficiency and effectiveness of all phases of the cycle. The PPAFO assists impaired gait by: (1) controlling the motion of the foot at heel strike, (2) providing modest assistive torque for propulsion and stability during stance, (3) supporting the foot in the neutral (90° joint angle) position during swing to prevent foot-drop, and (4) allowing free range-of-motion during the rest of the cycle (Fig. 1A). A 255 g (9 oz.) portable compressed liquid CO2 bottle and pressure regulator (JacPac J-6901-91; Pipeline Inc., Waterloo, Canada) are used to power a dual-vane bidirectional rotary actuator (CRB2BW40-90D-DIM00653; SMC Corp of America, Noblesville, IN, USA) located at the ankle joint (Fig. 1B). Two solenoid valves (VOVG 5V; Festo Corp, Hauppauge, NY) connect the power source to the actuator or vent the CO2 to atmosphere. The solenoid valves are either fully open or closed and cannot be used to modulate actuator torque. The pressure regulator on the CO2 bottle and a second regulator (LRMA-QS4; Festo Corp US, Hauppauge, NY) located on the PPAFO were used to fix the magnitude of the dorsiflexor (toes-up) and plantarflexor (toes-down) actuator torque. The structural shell consisted of a tibial section and foot plate, which were custom fabricated from pre-impregnated carbon composite laminate material. A free motion ankle hinge joint connected the foot plate and tibial section on the medial aspect. Velcro straps secure the PPAFO to the shank and foot. The direction of the PPAFO torque is switched from dorsiflexor (toes-up) to plantarflexor (toes-down) based on valve control. Control of the valves was accomplished through the use of two force sensors, and an angle sensor (force sensor: 2in 2in square; Interlink Electronics Inc., Camarillo, CA, USA; angle sensor: 53 Series; Honeywell, Golden Valley, MN, USA). Onboard electronics (eZ430-F2013 Development Tool; Texas Instruments, Dallas, TX, USA) were used to control the PPAFO. The use of onboard electronics and a portable power source enabled the PPAFO to provide untethered assistance. While the PPAFO in the above configuration successfully provides assistance during gait (please see reference [7] for details), this system has two previously mentioned limitations: (1) the inability to provide intermediate levels of torque assistance, and (2) excessive actuation and high pneumatic power consumption caused by the current control scheme. To address these issues, a second PPAFO hardware configuration incorporating a single high-speed proportional valve (LS-V05s; Enfield Technologies, Trumbull, CT, USA) in place of the two solenoid valves was considered. 2 Copyright © 2011 by ASME Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 09/19/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use Modeling the PPAFO The pneumatic system model consists of the proportional valve, rotary actuator, pneumatic lines connecting the valve to the actuator, and the added inertia of the PPAFO foot plate (Fig. 2). The solenoid valve was modeled as a fully open proportional valve with a different cross-sectional area. The inertia and damping of the actuator vane and PPAFO foot plate were modeled as a single rigid body. Additionally, the following assumptions and simplifications were made: constant pressure at the source, no leakage within the system (except for leakage across the actuator vane), homogeneous pressure inside each chamber, negligible gas inertia, isothermal processes, and negligible line losses between the valve and actuator. In order to determine the angular position of the PPAFO(θ), the dynamics of the system were modeled with the following relationship, zz gravity f ex actuator I T T T T θ βθ + + + + = ɺɺ ɺ (1) In (1), θ is the vane position angle (which corresponds to the ankle joint angle), Izz is the moment of inertia of the foot plate and actuator vane relative to the ankle joint axis of rotation, β is the rotary damping ratio due to friction, Tgravity is the gravitational torque due to the weight of the PPAFO, Tex represents the coupling torque between the PPAFO and the wearer, Tf is the frictional torque opposing the motion of the vane, and Tactuator is the output torque created by the actuator (Fig. 2). The actuator torque was calculated using the following equation [10]: 1 2 ( ) actuator actuator T P P K = − (2) where Kactuator is an experimentally derived torque-to-pressure ratio for the rotary actuator, and Pi are the pressures in the two actuator chambers. The instantaneous pressure in chamber i was calculated using the ideal gas law,