Power Wheelchair Footplate Pressure and Positioning Sensor

Abstract

Power wheelchair users are at risk for severe injuries caused by foot mis-position on the footplate. This can lead to collisions or foot dragging which are severe or life-threatening injuries for people with spinal cord injuries. The foot cannot be safely immobilized due to tilting pressure relief injuries, therefore, the foot can easily fall into a vulnerable position without the user realizing it. To reduce the likelihood of injury, we have developed a sensor for monitoring foot position in real time, as the wheelchair is driven. The sensor uses an array of force-sensing resistors and infrared distance sensors to detect the pressure and location of the foot within the immediate confines of the footplate. Sensor arrays with 23 force sensors and 14 infrared sensors per foot were fabricated on standard printed circuit boards and encapsulated in a durable thermoplastic urethane for environmental resistance. Fabricated sensors transmitted foot pressures and position data at 10 Hz using a Bluetooth Low Energy radio. An iOS app was developed to notify users of vulnerable foot position. Measured results confirmed the functionality of the system over typical foot pressures, and indicated that the device is ready for next-stage clinical trials with spinal cord injured power wheelchair users.

I. Introduction

Complex Rehabilitation Powerchairs (PWCs) are critically important for people with spinal cord injury and disorders (SCI/D), providing mobility independence and improved quality of life. However, PWCs can be dangerous for users who cannot easily feel or reposition their lower limbs. Many conditions and activities of daily living, e.g. driving over rough terrain, can displace the feet from the footplates (Fig. 1). Essential pressure relief using tilt-in-space regularly causes foot displacement as the footrests extend as the chair tilts backward. Further, uncontrolled limb movements due to spasticity can cause limbs to be mispositioned without user sensation, and if individuals are also obese they might not visualize their limb positions.

Figure 1.

Power wheelchair systems include a footplate (A) which does not always prevent a user’s feet from contacting the ground or nearby objects, depending on common postures (B,C).

When the feet are mispositioned, injuries can occur due to pressure points and the foot dragging or colliding with obstacles. PWC-related injuries include fractures and lacerations, with 6.7–33% of injuries caused by catching a lower extremity on a doorframe during wheelchair use [1–2] [3]. Due to greatly reduced bone strength in SCI/D people, tibial and femoral fractures occur even in low speed collisions [4–5]. These devastating injuries often lead to prolonged hospitalization and have even caused death.

The foot cannot be safely immobilized for most SCI/D PWC users due to extensor spasms or foot movement during pressure relief tilting. Therefore, the foot can easily be mispositioned on the footplate, and in a vulnerable position, without the user realizing it. We propose that a Footplate Pressure and Positioning Sensor (FoPPS) can be used to determine the foot’s position in real-time during PWC use. The FoPPS could warn the user if the feet leave the safe confines of the footplates, or could communicate with the PWC control systems to slow or halt the chair to prevent injury. Here we present, for the first time, the design and testing of a FoPPS prototype for use on PWC footplates and alerting users of mispositioned feet via a smartphone app.

II. Footplate Pressure and Positioning Sensor

A. Sensor Architecture

Foot position could be detected relative to the footplate with many sensor modalities. To accommodate a wide range of footwear, be cost-effective, and increase the environmental resilience of FoPPS, we adopted a simple architecture consisting of an array of force-sensing resistors and infrared distance sensors (Fig. 3). Similar systems have been used for determining pressures or forces in prosthetics [6–9]. This topology was chosen to be economical and extremely durable when overcoated with water- and abrasion-proof thermoplastic urethane (TPU) coatings. The force-sensing FoPPS has additional clinical utility, because it can determine if a user’s foot pressures are changing over time, which might indicate pressure ulcer risk or a poorly-fitted PWC. The IR sensors can determine if a person’s foot is frequently losing contact with the footplate during normal use, e.g. bouncing up and down but staying within the footplate area, to prevent false alarms based on loss of contact pressure.

Figure 3.

The output from the higher resolution CONFORMat system overlaid with the FoPPS sensor array (a), provided expected pressure values for design of the three variations of interdigitated force sensors (b).

B. Footplate-scale sensor arrays

The FoPPS IR sensors and FSRs are arranged on a grid across the footplate. To aid in the design of FSRs capable of sensing the full range of expected pressures, volunteers were seated upright in a chair with their feet resting on a CONFORMat (Tekscan Inc., Boston, MA) pressure sensor mat, which comprises a 14.7 mm² array of over 1,000 sensels. The highest pressures were observed under the heel at 1.3psi. Less pressure was observed under the forefoot, and little, if any, pressure was observed under the midfoot. Expected pressure values were adjusted by a safety factor of 150% to accommodate heavier users.

To determine the pressure difference between the heel, midfoot, and forefoot, the FoPPS uses three FSR types (Fig. 4). To balance refresh rate and cost with pressure map accuracy, the FSRs on the FoPPS were 30 mm squares, spaced 10 mm between edges. This sensor density still provides ample resolution to find high- or low-pressure areas when compared to the higher-resolution CONFORMat sensor (Fig. 4). The IR sensors used to infer distance from the footplate are placed in rows between the FSRs (Fig. 2).

Figure 4.

The sensor arrays use a common row-column addressing for individual FSR measurement, with all IR phototransistor collectors sharing a common readout bus (a). IR sensor readout uses an AC-coupled charge decay scheme (b). A correlated dual measurement is made at each array site beginning with the IR LED turned off to measure the dark-state offset.

Figure 2.

The FoPPS architecture uses an array of interdigitated force sensors and IR distance sensors to determine shoe position and pressure on the footplate. IR sensors can sense regions of the shoe that are close to the footplate but not touching it, e.g. a mid-sole arch common in shoes.

C. Sensor array readout electronics

FoPPS sensors are read out with a row-column multiplexed scheme (Fig. 4a). All FSRs are measured with a simple voltage divider using a reference resistor. A ratiometric measurement using a 10-bit analog-to-digital converter (ADC) is used.

IR sensors are measured using a correlated double-sampling, AC-coupled approach to improve rejection of ambient IR light. To reduce the number of multiplexers, all IR detectors share a common bus; the bus photocurrent I_PC is thus determined from the total IR light striking the whole sensor array. Measurements of each array location are performed using a charge-decay scheme in which a holding capacitor is precharged, released, and the time delay of charge leakage is measured. A time-to-digital conversion is performed by measuring the time taken for the IR bus voltage V_IR to cross the logic threshold V_L. This topology is easily implemented with high-speed digital microcontrollers (Fig. 4b).

To determine the value of an individual site, a baseline measurement of the IR bus voltage is made with all IR LEDs off. Immediately after this measurement, a single IR LED is illuminated and another measurement is made. The difference between the two measurements represents the total light reflected to a specific sensor site.

FSR and IR signals from the left and right footplates are multiplexed onto separate channels, so a total of 2 ADC inputs and 2 general purpose input-output pins are needed to capture array values. A set of quantized sensor values for each sensor site forms a sensor “frame” with each frame comprising N_FSR=23 & N_IR=14 sites per foot. IR sensors are measured twice in each conversion for correlated double sampling.

For typical microcontrollers, a frame can be produced within 10 ms with a multiplexer update rate of 3.7 kHz. Assuming 8-bit truncation and a frame rate of 10 frames per second, the FoPPS data rate is low at only 370 bytes/sec. This data rate can easily be handled by typical low-power wireless radios, e.g. Bluetooth Low Energy.

D. Sensor fabrication

The FoPPS system was fabricated on a commercial fiberglass PCB to maintain low cost. FSRs were created on the top metal with the geometry in Table I. An electroless nickel, immersion gold finish was used to minimize contact resistance on the FSR traces. IR sensors, multiplexers, and passive components were all mounted to the PCB. The total assembled electronics thickness was less than 3 mm. Wire header connectors were used to connect the left and right footplates together to a RedBear Duo (RedBear Lab, Hong Kong, CN) Arduino-compatible board (Fig. 5). The Duo strobed bilaterally through all sensor multiplexers, using an onboard ADC for sensor conversion. The sensor strobe rate was programmed to produce a full “frame” of both sides at 10 frames per second. An onboard Bluetooth Low Energy (BLE) 4.1 radio was used to transmit sensor frames as they were received, i.e. 10 times per second. The radio had an effective line-of-sight range of 20 meters, and the Duo consumed 35 mA continuously while measuring and transmitting FoPPS data. The system was powered by a 500 mAh lithium polymer secondary cell, which would allow 2 weeks of continuous monitoring between recharges.

TABLE I.

Interdigitated FSR Trace Dimensions

	Dimension A	Dimension B
Fine FSRs	0.152mm	0.152mm
Medium FSRs	0.203mm	0.203mm
Coarse FSRs	0.254mm	0.254mm

Figure 5.

Fabricated and encapsulated FoPPS devices, sized for use on standard PWC footplates, and with real-time data streaming to a smartphone for user notification of foot mis-position.

The FoPPS electronics were overcoated in a multi-layer process to preserve FSR and IR function while adding environmental and abrasion resistance. A 2-piece protective TPU overcoating was 3D-printed on a low-cost fused filament fabrication (FFF) printer. Small 1” squares of 2mm XactFSR film (Sensitronics, Bow, WA) were glued to the underside of the top layer of the TPU overcoating such that they rested directly above each FSR on the FoPPS. A 2mm tall bump in the top of the TPU overcoating was positioned above each FSR on the FoPPS to act as a stress concentrator, ensuring that pressure from a patient’s foot would be distributed onto an FSR, and not another part of the FoPPS PCB. A silicone caulk glued the two TPU parts together, fitting snuggly over the PCB and completely encapsulating it (Fig. 5).

III. Sensor Characterization and Demonstration

Fabricated prototypes were first characterized in static conditions, and then tested with a prosthetic lower limb extremity outfitted with an athletic shoe to represent realistic force distributions. Testing with the prosthetic limb generated real-time readouts of FoPPS data via the smartphone app; this data is presented qualitatively.

A. Force sensor response curve

Each of the three FSR layouts (coarse, fine, and medium) was tested using precision masses (0.00–1.05 kg) which were applied successively to a single sensor site. This range was selected based on pressure ranges attained from the CONFORMat data with human volunteers.

Static resistance measurements were made with an Agilent 34410A digital multimeter. All sensors showed a monotonic, nonlinear response to applied force (Fig. 6). As expected, the fine sensor was the most sensitive to applied force. None of the sensors saturated, although all sensors had reduced sensitivity above 200g loading. This suggests that sensor overload is unlikely to occur with typical foot pressure distributions and this FSR topology, even with the finest FSR spacing.

Figure 6.

The three variations in the FSRs used by the FoPPS showed monotonic response curves to weight, sufficient to determine foot pressure distribution across the PWC footplate.

B. Infrared sensor response curve

The IR sensor distance response was tested using a black rubber target to simulate the underside of a shoe. IR sensors were pulsed as described previously to reduce ambient light source interference. The Duo microcontroller was programmed to produce the switching and correlated charge decay measurements described in Fig. 5b. Component values were selected to attain a nominal τ_D of 500 μs, and the Duo could calculate delay time within approximately 1% based on code executing speed limits. The IR LED was controlled by the Due using a current limiting resistor to set the peak current to 20 mA. IR distance data were transmitted to a computer serial port for data collection. Lighting conditions in this test were approximately constant, although the IR sensor did not display sensitivity to ambient lighting conditions, as designed. A typical IR sensor response curve showed a monotonic response to target distance, with greatly reduced sensitivity beyond 2 cm (Fig. 7). This range was sufficient for foot detection and could be tuned at the expense of IR LED power consumption.

Figure 7.

Measured response of the IR sensors. The sensitivity drops after 2 cm which improves detection of excessive foot-footplate spacing

C. Wireless readout electronics and smartphone app

Sensor data were received by a custom iOS app running on an iPhone 6 or an iPod Touch. This app was developed to simulate real-time detection of foot position to provide feedback and warnings to users. The app has three user interfaces (Fig. 8) designed to be simple to interpret for PWC users who might be distracted by activities of daily living. The app receives sensor frames and processes them to determine if the user should be alerted of a dangerous foot position. The app was programmed with a simple algorithm which averaged the footplate pressure and alerted the user if the weighted sensor values were below a threshold. The three user interfaces are shown below (Fig. 8). Overall, the performance of this system was sufficient for pilot trials (Table II).

Figure 8.

FoPPS app output with prosthetic shoe on left footplate. The app includes three screens for user notification: overall status (a), a live pressure + IR distance map (b), and a visual status for each footplate (c).

TABLE II.

Performance Summary

Sensor size	145 × 206 cm per foot
FSR sensors	46 (23 per foot)
IR sensors	28 (14 per foot)
Force sensing range	0 – 1150 g minimum
IR Distance sensing range	2.0 cm
Wireless sensor frame rate	10 frames per second
Wireless protocol	Bluetooth Low Energy
Power draw	35 mA from 3.6-V battery

IV. Conclusion

We have developed a prototype foot position sensing system capable of alerting a PWC user if their foot is in a dangerous position. The FoPPS system was designed with a simple topology for readout using common microcontroller platforms. Force and infrared distance sensors produce a real-time map of foot location, footplate pressure, and foot distance from the footplate at frame rates of 10 Hz. This data was sufficient to detect real-time mispositions of the foot and communicate alerts to a smartphone app. We are currently testing the FoPPS with PWC users to determine natural foot movement during activities of daily living and to evaluate user satisfaction with the FoPPS alter interface options.