1. Introduction
In Manual Wheelchair (MWC) sports, the optimization of mechanical parts can affect performance (Haydon et al. Citation2018). As mechanical power is dissipated by the MWC rolling resistance (RRes), the user performance could be enhanced by the optimization of this factor. RRes depends on wheel, floor covering and MWC setting as well as user positioning (Bascou et al. Citation2013). Its estimation is not simple and the athlete’s perception may not be precise enough to assess it, especially because of ground irregularities. However, RRes could be estimated using 3 D accelerometers. Many smartphones are provided with 3D accelerometer sensors and could then be used as a massive, easy way to estimated RRes, providing wheelchair sport trainers and therapists access to this measurement. However, the sensors used in the literature are specialized and precise devices. On the contrary, conventional smartphones may not have enough precision to allow correct estimation of RRes, as they do in other applications (Dietz et al. Citation2017).
This study aims at verifying to which extent common smartphones can be used to assess RRes in the field.
2. Methods
2.1. Material and protocol
4 smartphones equipped with 3D accelerometers (‘ASP’) were used in this study: Samsung Galaxy A3 (ASP 1), Galaxy A5 (ASP 2), Galaxy S7 (ASP 3) and Sony Xperia M3 (ASP 5). ‘Physical tools’ free application was used to measure accelerations with the ASP. The decelerations they measured were compared to 2 inertial measurement units (‘IMU’, MT Awinda, Xsens). Their sensibility and offset were controlled using an electronic spirit level: a perfectly horizontal line was set on the wall using the spirit level, and each smartphone was placed on it, aligning the smartphone x, then y, −x and − y axis with the gravity. Z and − z axis offset and sensibility correction was achieved the same way, using a horizontal plane.
Deceleration tests were conducted on two different floor coverings (‘carpet’/ ‘vinyl’), for two MWC (‘1’: Otto bock voyager and ‘2’: Vermeiren Eclips), loaded with 40Kg additional masses, placed on two positions (‘front’/ ‘rear’), resulting in 8 MWC settings (). All calibrated ASPs and IMUs were fixed with double-sided tape on a wooden plate, which was attached using a belt to the MWC, above the additional masses. The x axis of ASP and IMU were aligned manually with the movement direction.
Table 1. MWC setting conditions.
For each MWC setting, 10 free deceleration tests were performed in both fore-and-aft directions. The protocol was 1) 2 sec static phase on a departure mark fixed on the ground, 2) clean manual push to accelerate the MWC between two 1 m-separated marks, 3) free deceleration while verifying the straightness of the MWC path, 4) clean manual stop between two ending marks, 5) 2 sec static phase. Deceleration tests and data treatment were conducted in accordance with Bascou et al. (Citation2013).
10 more trials were conducted for the first MWC condition (C1a), changing only the ASP and IMU positions on the wooden plate (C1b, see ).
2.2. Data treatment
Static phases were used to correct the ASP and IMU misalignments. Free deceleration phase was automatically extracted using the clean accelerating and stopping phases. Deceleration value for the trial was the mean of the x-acceleration measured during the deceleration phase, removing the residual deceleration measured during the ending static phase. Fore-and-aft values for 1 trial were averaged to correct the ground inclination between start and stop points, conducing to produce one deceleration value per trial.
Mean, standard deviations of the ASP were compared to IMU ones. Non-parametric, paired Wilcoxon-Mann-Whitney tests were used to compare ASPs and IMUs.
3. Results and discussion
The decelerations measured ranged from 0.09 to 0.25 m/s2, depending on the flooring, MWC conditions and ASP/IMU. Decelerations varied from 3 to 9% within the same set, justifying conducting various trials per set to achieve a correct assessment of the deceleration value.
Mean absolute difference between IMU decelerations values was 1.6%, which were considered consistent with one another. The average value of the IMU decelerations was taken to compare to ASP deceleration values. Averaging all series for each ASP, the differences with IMU ranged between 9% for ASP 1 and 2 to 15% for ASP 3 and 4. This difference could reach up to 28% considering the series and ASP individually.
Deceleration errors were reduced for carpet floor (8%) error) or additional mass positioned forward (10% error), compared to vynil floor (16% error) or additional mass backward (14% error) conditions. This suggests that the greater deceleration measured, the lower the error, as carpet ground and forward mass increased RRes and deceleration.
Mixing the values of every ASP or considering each ASP individually, Wilcoxon-Mann-Whitney tests did not show statistical difference with IMUs.
ASP programs, memory, calculation capacity and age seemed to affect the results, especially for ASP3, which experienced freezing during the measurements and showed the lowest precision.
However, every ASP was able to rank correctly the C1 to C7 conditions, according to their deceleration, except for C3 and C8 conditions, which had very similar decelerations (0.19 and 0.18 m/s2 respectively).
4. Conclusions
In this study only 4 smartphones were tested, but the results seem promising, as they all were able to differentiate between MWC settings and to achieve sufficient precision to rank them by their decelerations. The precision may increase in the future as the power of the smartphones gradually rises.
Allowing the simple measurement of RRes, in the field, for any MWC, using personal smartphones may allow increasing the use of objective data when choosing MWC settings. An accessible tool, allowing the easy data treatment may allow athletes, trainers and occupational therapist to include RRes measurement in their daily use.
Figure 1. MWC 1, with front load, ASPs and IMUs.
References
- Bascou J, Sauret C, Pillet H, Vaslin P, Thoreux P, Lavaste F. 2013. A method for the field assessment of rolling resistance properties of manual wheelchairs. Computer Method Biomech Biomed Eng. 16(4):381–391.
- Dietz MJ, Sprando D, Hanselman AE, Regier MD, Frye BM. 2017. Smartphone assessment of knee flexion compared to radiographic standards. Knee. 24(2):224–230.
- Haydon DS, Pinder RA, Grinxhaw PN, Robertson W. 2018. Using a robust design approach to optimize chair set-up in wheelchair sport. Proceedings of ISEA 2018, March 26, Brisbane (AUS).