The design of the Perturbation Platform System was guided primarily by the goal to operate the device remotely by a single individual and reset to the start position on command while supporting the participant’s body weight. The system was designed to provide a sudden vertical displacement of the standing support surface up to 76.2 mm and provide either unilateral or bilateral perturbations. This distance was selected because it is sufficient to trigger a rapid postural reaction for balance recovery and the impact force is reasonable (approximately 1 bodyweight) for safe, repeated exposures . The platform can also tilt up to 15° in both directions providing the ability to perform ankle inversion/eversion or plantarflexion/dorsiflexion perturbations depending on orientation. These angle magnitudes have been established to provide a maximum perturbation without risk of injury . In addition, the platform was designed to be triggered on foot contact from the user, allowing the direction and distance of the drop to be concealed to simulate unexpected perturbations. Lastly, the system is relatively portable such that it can be moved by a single individual for setup in a gait laboratory or used in a clinical setting. To this end, the main components of the system were constructed from ABS plastic with 2 platform units each weighing 25 kgs, measuring 406 mm wide x 508 mm long x 236 mm tall and include handles for ease of transport (Fig. 1). Final implementation in the laboratory setting includes in-ground force plates directly beneath each platform and an elevated stage surrounding the system to facilitate overground walking trials as well as helping to alleviate potential anxiety due to the sensation of height when standing on the platforms. The system is described in detail below.
The Perturbation Platform System consists of three main components: 1) two movable standing surfaces, 2) a high-pressure air source, and 3) a remote control interface (Fig. 1). The operation of each platform utilizes four, double acting, pneumatic pistons (Space Saver Low Profile, SS-150, Mead Fluid Dynamics, Chicago, IL, USA) (Fig. 2A). Each piston features 76.2 mm of travel with the capability to lift 80 kg at a working pressure of 6.9 Bar. With four pistons per platform, each platform can lift a 320 kg person or 640 kg when two systems (i.e., bilateral perturbations) are used together. The system is controlled via a 5/3 directional air valve (IMI Norgren, K81EA00KV0KV02W1, Littleton, CO, USA) (Fig. 2B). The valve is controlled with two 12-volt solenoids that can be activated or deactivated to raise or lower the pistons. The use of the 5/3 valve allows for directional control of the pistons along with a third mode where airflow is closed, which is activated when no electrical signal is applied. This provides a safety feature in the event of power loss that will lock the pistons in place without dropping the subject unexpectedly. A single, high-pressure air source is provided from an external compressor (Fig. 2C). Currently, the compressor source is in a separate room with the air line routed through the floor to provide quiet and discreet operation of the platforms. Air is supplied and exhausted from the pistons evenly via two distribution blocks (Fig. 2D). An exhaust muffler helps to quiet exiting air to prevent startling participants or alerting them to the timing of upcoming perturbations (Fig. 2E). Lastly, each piston is fitted with a one-way variable flow control valve at the inlet/outlet to allow for speed control of the piston drop and return.
Movement of the upper platform is controlled via four, 12-volt, sealed linear solenoids (Magnet Shultz of America, S-07791, Westmont, IL, USA) (Fig. 3A). The linear movement of the top plate is then guided up and down on four V-Groove track roller and rail elements (Fig. 3B and 3C, respectively). The linear solenoids are threaded into brass inserts (Fig. 3D) that are pressed into the moving top plate and contain a hardened steel pin. When the 12-volt signal is applied the pin is retracted into the solenoid. When the signal is removed an internal spring pushes the pin outward to mate with a corresponding brass insert (Fig. 3E) on the two end plates of the platform box. The solenoids and pins allow for the selection of different drop types (straight vs angle drop). When all pins are retracted the plate is guided by the track rollers and drops straight down. When a single pair of pins is engaged on opposite ends of the platform the top plate rotates around the axis created by the pins to enable angled orientations in both directions (i.e., inversion/eversion or plantarflexion/dorsiflexion). During angled operation, the track rollers lose contact and the motion of the top plate is guided solely by the pins. As a safety precaution, in the event of a loss of power to the solenoids, all pins are engaged resulting in the top plate being locked in the up position. To enable different levels of straight and angle drops, adjustable stops are used to provide an end position for the top plate after a drop (Fig. 2F). Six different positions are provided via predrilled holes in the end plates of the platforms. The outermost holes facilitate 25.4 mm, 50.8 mm and 76.2 mm straight drop heights while the inner set of holes allows for 5°, 10° and 15° drop angles. Lastly, to enable the top plate to drop on contact with the user, four spring pins are located just under the top plate at the corners, two in each end plate (Fig. 4). The edges of the top plate are lined with a strip of hardened steel such that the impact with the pin does not deform the plastic material. The engagement of the spring pins can be varied by moving the tips in or out in their threaded mounts in the end plates that allow for adjustments of the force magnitude required to cause them to fall.
The controller for the system utilizes a Teensy 3.2 microcontroller (PRJC Sherwood, OR, USA) (Fig. 5). The controller is hard wired to the two platforms with connectors at each end to allow for assembly and disassembly. The 4-meter lead allows for remote operation of the system by a single person from a central location in the laboratory. The controller consists of four inputs. The two on either side allow users to set the drop mode of each platform individually and includes an LED indicator to indicate the current mode of operation (straight drop, inversion/plantarflexion drop, eversion/dorsiflexion drop or all lock). The two vertical buttons in the center initiate the drop or return for both platforms simultaneously for simple activation of each test condition. Varying modes of operation can be accomplished with the two platforms combined (Fig. 6), which allows a wide range of unilateral or bilateral testing scenarios.
Testing was performed to verify the function and capability of the Perturbation Platform system. A single subject (male, 1.66 meters, 73 kgs, 36 years old) provided informed consent to the testing as approved by the University of Texas Internal Review Board. While standing with one foot on each platform, the system was run through the nine configurations available (Fig. 6) and returned to the original position with the subject remaining on the platform.
In order to characterize the motion of the system, a Vicon motion capture system (Vicon, Centennial, CO, USA) was used to track the motion of a marker attached on the top plate of the platform to quantify the accelerations experienced over repeated drops. With the individual standing with one foot on each platform, five, unilateral, straight drops were performed at each drop height (Min: 25.4 mm, Mid: 50.8 mm, and Max: 76.2 mm) for a total of 15 drops. To characterize the motion of the platform the instantaneous velocity and acceleration were calculated by taking the first and second derivatives of the measured marker position over time. The derivative signals were then smoothed using a 6 Hz, lowpass, Butterworth filter and the peak downward acceleration was determined. Results from the three drop heights were then compared using a one-way ANOVA and multiple comparison test to determine overall and pairwise differences (p < 0.05) between the drop heights, respectively. All statistical testing performed in Matlab (MathWorks, Natick, MA, USA).