Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing for Injury Prevention based on Inertial Sensor Data

Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing... Current Directions in Biomedical Engineering 2019; 5(1):183-186 Michael Munz*, Thomas Engleder Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing for Injury Prevention based on Inertial Sensor Data Abstract: In this work, an assistant system is presented for climbing gym in 2012 recorded a total of 161 accidents the automatic assessment of falling and belaying in sport involving serious injuries (ambulance use) in 31 climbing climbing. Both climber and belayer are equipped with inertial gyms. In this study, the behaviour during belaying was also measurement unit (IMU) sensors. Forces as well as examined and an average of 1.4 belaying errors per lead movements in the form of multi-dimensional accelerations on climb and 0.7 errors per top-rope [1] were recorded. the legs and torso are captured. It can be shown that forces However, this doesn’t mean that every belay error can be estimated by means of IMU sensors, thus eliminating automatically leads to an accident. Usually a combination of a complex force measurement unit in the safety chain. errors and unfortunate circumstances is necessary, such as the Furthermore, the data can be used to assess both falling and release of the rope with simultaneous fall of the climber, a belaying by automatic segmentation and evaluation too long jump off the fall and subsequent striking on a algorithms. The sensor data should later be evaluated volume, etc. Faulty belaying thus rarely leads to an accident automatically in order to objectively measure faulty behavior and therefore nearly no feedback to the belayer is provided. by climber or belayer (for example wrong jump-off behavior, This can lead to a dangerous adaptation. One possibility to too hard protection, etc.). The overall goal is to provide realize injury prevention is specialized fall training, which quantified feedback in fall training for injury and accident individually improves the technique and reduces the prevention. frequency of mistakes. However, mistakes in belaying are independent of climbing ability: the climbing experience or Keywords: Accident prevention, sports climbing, fall climbing difficulty does not correlate with the quality of analysis, wireless body sensor networks, inertial belaying. Therefore, even experienced climbers are not measurement units protected from belay errors. Fall and safety training should regularly be practiced not only by beginners, but also by https://doi.org/10.1515/cdbme-2019-0047 experienced climbers. The basic idea of the fall and safety instructor presented in this work provides a starting point for training evaluation and objective feedback of the quality to 1 Introduction both climber and belayer. For this, both belayer and climber are equipped with inertial measurement unit (IMU) sensors. In recent years, sports climbing gained enormous The transmission of the measured data is done with popularity and is practiced not only on rock but also in Bluetooth, the evaluation is possible on a notebook, tablet or climbing gyms. A climbing rope is used for belaying the even smartphone. The system can be used universally in climber, which is attached to multiple anchor in the wall every route and with every backup device. Training feedback using quickdraws (a combination of slings and carabiners). can be used to monitor continuous improvement. A similar However, studies show a large number of safety failures, sensor system was presented in [4], but only climbing falls which are the main cause of accidents in climbing gyms were detected, no further assessment of quality aspects was [1,2]. Many injuries are most likely caused by over- provided. tightening, e.g. hard belaying, resulting into strong impacts One of the most important factors of a good belaying into the wall [3]. The German Alpine Club DAV study about technique is the distinguishing between dynamic and hard belaying. Soft belaying is achieved when the belayer moves towards the wall upon breaking. Hard belaying is achieved ______ when the belayer either stays passive in his position or even *Corresponding author: Michael Munz: University of Applied Sciences Ulm, Germany, Michael.munz@thu.de moves away from the wall. Although prolongating the fall Thomas Engleder: University of Applied Sciences Ulm, Germany Open Access. © 2019 Michael Munz et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution-NonCommercial- NoDerivatives 4.0 License. M. Munz et al., Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing for Injury Prevention based Se nso r D ata — 184 on Inertial length, dynamic belaying reduces the maximum deceleration 2.2 Execution of Trials force. Hard belaying introduces additional force into the Several trials were executed in order to prove the basic system, resulting in a strong impact to the climber towards feasibility of metrological acquisition of the required the wall, which often leads to injuries of the climber. quantities by means of IMUs. Forces are evaluated by means Therefore, the dynamic of the belay technique is important to of force measuring sensors. However, this should later be prevent injuries [5]. completely replaced by IMU measurements. The quantities to be acquired by the system are: In total, 16 trials with different conditions were carried out in  Forces on the climber a climbing gym during toprope and lead falls (see table 1). In  Fall duration, fall distance toprope, some of the falls were deliberately soft or very  Accelerations on the legs of the climber during impact dynamic, some passive or even with hard belay. The fall  Trajectory of the climber during fall (rotation, etc.) lengths in toprope situation were about 2 m each. This  Movement of the belayer (upwards, downwards, etc.) corresponds to the amount of slack rope deliberately issued by the belayer. In the lead, the last quickdraw was exceeded Sensors can determine these quantities metrologically. From by 0.7 m to 1.5 m, resulting in fall lengths of 1.5 m to 3 m. the measured data, quality criteria (scores) for the fall and the IMU sensors were attached to the body's centre of gravity belay process are to be calculated and evaluated in a next (COG) of both climber and belayer and to the climber's step. These are among others: ankles. The attachment of the sensor centrally on the back of  Strength of the impact, dynamics of belaying the waist belt is convenient to map the attack of the  Adequacy of the amount of slack rope gravitational force at the centre of mass. The acceleration  Jumping behaviour of the climber (too far away from the sensors on the ankles should provide additional information wall, rotation, etc.) about the impact energy when contacting the wall. In  Impact on the wall and force reduction addition, the load cell was inserted into the safety chain at the  Stance position of the belayer climber’s central belay loop at the harness. In this paper, we  Reaction of the belayer (passive / active, etc.) only present climber’s data. The scores should provide an objective measure of the In a climbing gym, DIN 2015 / 2015-07-00 defines the quality of the fall course, broken down into climbers and maximum distance between two quickdraws with height > belayer aspects. 8m above ground to 1.50 m. The trial conditions in table 1 therefore represent the maximum expected fall height in climbing without unnecessary slack rope in the system. 2 Material and Methods Table 1: List of trial conditions 2.1 Measurement principle No Type Free rope Slack Height above length [m] rope last quickdraw Measurements include force, acceleration, orientation and [m] [m] short-term position, and their temporal change. To detect 1 Dynamic 15,0 2,0 - forces usually load cells are used, which are inserted into the 2 belay 15,0 2,0 - safety chain. However, this introduces an additional risk 3 Hard belay 15,0 2,0 - (breakage, wrong integration, etc.) and affects belayer and 4 15,0 2,0 - climber in their normal behaviour. Therefore, in this work we 5 Wide jump 15,0 2,0 - 6 15,0 2,0 - show that forces can be reconstructed from the IMU data. To 7 “Normal” 12,0 2,0 1,0 validate this method, a load cell with a measuring range of ± 8 conditions 12,0 - 1,0 10 kN was used. 9 12,0 - 1,0 To measure the movement and accelerations, IMU 10 Leading 6,5 - 0,7 sensors (MPU-9150, InvenSense) were attached to climber climb with 11 8,0 - 0,7 and belayer. Each IMU consists of three-axis acceleration, different 12 9,0 - 1,5 gyro and magnetic field sensors. All data is digitally recorded conditions 13 10,0 - 0,7 14 11,0 - 1,5 by a microcontroller and transmitted wirelessly via Bluetooth 15 11,0 - 0,7 to a computer, allowing synchronous data acquisition of 16 12,0 - 1,5 several IMUs. The measurement rate was 100 Hz. M. Munz et al., Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing for Injury Prevention S ensor Data — 185 based on Inertial the integration is realized using a numerical integration method. The initial impulse of the fall is fit using linear regression, which corresponds to an unrestrained fall. The greatest deviation of the impulse from the regression line is at impact into the rope, which leads to the braking timestamp. Subsequently, the beginning and end of the fall are detected. For this purpose, a resting position is detected for the end of the fall, when only gravitational acceleration is measured for a certain period of time (after the decay phase). The start of the free fall is detected by comparing to a low threshold (additional acceleration due to the rope friction and the redundant belay needed for the trials). By this, the duration of the free fall is defined. The fall length can be calculated by double-integrating the acceleration. This is applicable because of the short integration time and the low acceleration noise. With the known body weight m of the climber, the force at the central belay loop of the climber can be estimated from the resulting acceleration using . The timestamps of the maxima during the deceleration Figure 1: Resulting acceleration (yellow) and segmented process (impact force) in both and measured force are phases / events during fall: jump start (blue), jump end determined. The time offset and the difference between the (magenta), max. acceleration (green) measured at the COG of the climber. Top: hard belay, bottom: dynamic maximum values serve as a measure of the correspondence belay. between the estimated value and the real force measurement . 2.3 Data processing From the data of the inertial sensors it is possible to deduce the orientation of the sensor at any time. For this, a gradient 3 Results and discussion decent algorithm [6] for fusing acceleration and gyro The evaluation of the force measurements shows a temporal measurements of the IMU sensors is used. In addition to the deviation of 0.01 s (median) with a range of -0.04 to +0.03 s accelerations, this also allows the body movements and body (1st or 3rd quartile) over all measurements. The duration of position to be determined, which provide essential information for the later scores. the impact peak according to [7] is between 0.1 s and 0.2 s. Therefore, the results are within the time window of the The evaluation of the measurement results focuses primarily on the measurement data of the force sensor and impact peak, and so the maximum impact was always detected correctly. the acceleration values at the COG. Supplementary information is obtained by the determined acceleration values For the deviation of the estimated force from the at the ankles. The coordinate axes of the acceleration sensors measured force , the following values were obtained for all span a Cartesian coordinate system. The resulting absolute experiments: Median: -112 N, range: -512 N to -102 N (1st acceleration can be calculated based on the accelerations and 3rd quartile, respectively). In addition to the force peaks, along the x-, y- and z-axis using = the course of the estimated and measured force was also examined. For this purpose, the course from the beginning of Based on , the fall can be segmented into different the deceleration process to the capture impact was subjected phases, as shown in figure 1. Start and end of the fall, time to a congruence analysis. Figure 2 shows results of the and strength of the maximum impact into the rope are individual trials as box plots. All values are normalized to the automatically calculated by a segmentation algorithm. For maximum observed force of 2050 N. this purpose, various filters and pattern recognition The difference during the phase of the free fall is due to techniques are applied to the measurement data. the rope friction in the carabiners of the quickdraws, leading First, the data of the acceleration sensor is low pass to greater differences in trials with longer falls. filtered for noise reduction. Afterwards, the impulse is Analysing the acceleration data in more detail, additional estimated from the acceleration data using , where assessment can be conducted. The strength of the impact M. Munz et al., Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing for Injury Prevention based on Inertial Sensor Data — 186 Figure 3: Example of a fall with asynchronous impact of feet Figure 2: Boxplot of the difference between measured force F after the main deceleration peak of the rope impact. and FB for all 16 experiments. The deviation between F and (green). is the amount of slack in a fall. This aspect must be assessed FB normalized to 2050 N is indicated on the y axis. in relation to the climbing height. Sensory extensions are planned, which will be evaluated in subsequent work, peaks allows the differentiation between soft and hard allowing an evaluation of a possible fall situation at any time. belaying. Unnecessarily strong initial take-off of the climber Further work has been done in developing algorithms for can also be detected using acceleration data and the direction scoring relevant quality criteria for safer and climbers on the of acceleration. By measuring the accelerations on the ankles, presented features extracted from acceleration data. The the delay between impact of both feet on the wall can be output of this scoring algorithm can then be evaluated as part detected. This serves as an indication of an unstable posture of a series of measurements with independent climbing and thus a poor control of the fall. Figure 3 shows a coaches. corresponding example: after the impact and breaking start in the rope, the feet touch the wall asynchronously (two peaks). Author Statement In addition, the strength of the impact of feet on the wall can Research funding: The authors state no funding involved. be determined. The shape of this peak can also serve as a Conflict of interest: Authors state no conflict of interest. feature for assessment of the impact attenuation by the climber’s feet. References 4 Conclusion and Outlook [1] Hummel, C. & Hellberg, F. „Unfälle an künstlichen Kletterhallen: Kampf dem Eisberg!“. DAV Panorama: Magazin des Deutschen Alpenvereins 6/2014, 62–65. In this work, a prototype of a fall and safety instructor for the [2] Schöffl, V.R., Hoffmann, G. & Küpper, T. (2013): Acute injury quantitative assessment of forces and movements in sport risk and severity in indoor climbing-a prospective analysis of climbing falls is presented, providing objective feedback to 515,337 indoor climbing wall visits in 5 years. – Wilderness & the athletes or coach. To the best of the authors’ knowledge, environmental medicine 24, 3, 187–194. [3] Radelzhofer, P. (2014500): Bergunfallstatistik 2012-2013. this is the very first system that can perform such an München. assessment in sports climbing for injury prevention. [4] Tonoli (2011): Fall identification in rock climbing using Validation measurements show that the impact force at wearable device. In: Journal of Animal Science (X), S. 1–11. DOI: the climber's rope point can be estimated using IMU 10.1177/ToBeAssigned. [5] Schöffl, Volker; Morrison, Audry; Schöffl, Isabelle; Küpper, measurement data, allowing to quantify the softness of the Thomas (2012): The epidemiology of injury in mountaineering, belaying technique, as one important factor for injury rock and ice climbing. In: Medicine and sport science 58, S. 17– prevention. With a first analytical evaluation of the IMU 43. DOI: 10.1159/000338575. sensor data, we showed that it is possible to differentiate [6] Madgwick, S. O., Harrison, A. J., & Vaidyanathan, R. (2011): various processes and effects, allowing assessment of both “Estimation of IMU and MARG orientation using a gradient descent algorithm”. In Rehabilitation Robotics (ICORR), 2011 belay and fall quality. IEEE International Conference on (pp. 1-7). IEEE. A frequently discussed, very important parameter that [7] Fimml, W. & Larcher, M. (2000): „Energie ist Kraft mal Weg: cannot yet be measured and evaluated with the current Sicherungstheoretische Grundlagen, Teil 2“. bergundsteigen: system Zeitschrift für Risikomanagement im Bergsport 4/00, 14–20. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Current Directions in Biomedical Engineering de Gruyter

Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing for Injury Prevention based on Inertial Sensor Data

Download PDF
4 pages

Loading next page...
 
/lp/de-gruyter/intelligent-assistant-system-for-the-automatic-assessment-of-fall-97TMCC4cDw

References (6)

Publisher
de Gruyter
Copyright
© 2019 by Walter de Gruyter Berlin/Boston
eISSN
2364-5504
DOI
10.1515/cdbme-2019-0047
Publisher site
See Article on Publisher Site

Abstract

Current Directions in Biomedical Engineering 2019; 5(1):183-186 Michael Munz*, Thomas Engleder Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing for Injury Prevention based on Inertial Sensor Data Abstract: In this work, an assistant system is presented for climbing gym in 2012 recorded a total of 161 accidents the automatic assessment of falling and belaying in sport involving serious injuries (ambulance use) in 31 climbing climbing. Both climber and belayer are equipped with inertial gyms. In this study, the behaviour during belaying was also measurement unit (IMU) sensors. Forces as well as examined and an average of 1.4 belaying errors per lead movements in the form of multi-dimensional accelerations on climb and 0.7 errors per top-rope [1] were recorded. the legs and torso are captured. It can be shown that forces However, this doesn’t mean that every belay error can be estimated by means of IMU sensors, thus eliminating automatically leads to an accident. Usually a combination of a complex force measurement unit in the safety chain. errors and unfortunate circumstances is necessary, such as the Furthermore, the data can be used to assess both falling and release of the rope with simultaneous fall of the climber, a belaying by automatic segmentation and evaluation too long jump off the fall and subsequent striking on a algorithms. The sensor data should later be evaluated volume, etc. Faulty belaying thus rarely leads to an accident automatically in order to objectively measure faulty behavior and therefore nearly no feedback to the belayer is provided. by climber or belayer (for example wrong jump-off behavior, This can lead to a dangerous adaptation. One possibility to too hard protection, etc.). The overall goal is to provide realize injury prevention is specialized fall training, which quantified feedback in fall training for injury and accident individually improves the technique and reduces the prevention. frequency of mistakes. However, mistakes in belaying are independent of climbing ability: the climbing experience or Keywords: Accident prevention, sports climbing, fall climbing difficulty does not correlate with the quality of analysis, wireless body sensor networks, inertial belaying. Therefore, even experienced climbers are not measurement units protected from belay errors. Fall and safety training should regularly be practiced not only by beginners, but also by https://doi.org/10.1515/cdbme-2019-0047 experienced climbers. The basic idea of the fall and safety instructor presented in this work provides a starting point for training evaluation and objective feedback of the quality to 1 Introduction both climber and belayer. For this, both belayer and climber are equipped with inertial measurement unit (IMU) sensors. In recent years, sports climbing gained enormous The transmission of the measured data is done with popularity and is practiced not only on rock but also in Bluetooth, the evaluation is possible on a notebook, tablet or climbing gyms. A climbing rope is used for belaying the even smartphone. The system can be used universally in climber, which is attached to multiple anchor in the wall every route and with every backup device. Training feedback using quickdraws (a combination of slings and carabiners). can be used to monitor continuous improvement. A similar However, studies show a large number of safety failures, sensor system was presented in [4], but only climbing falls which are the main cause of accidents in climbing gyms were detected, no further assessment of quality aspects was [1,2]. Many injuries are most likely caused by over- provided. tightening, e.g. hard belaying, resulting into strong impacts One of the most important factors of a good belaying into the wall [3]. The German Alpine Club DAV study about technique is the distinguishing between dynamic and hard belaying. Soft belaying is achieved when the belayer moves towards the wall upon breaking. Hard belaying is achieved ______ when the belayer either stays passive in his position or even *Corresponding author: Michael Munz: University of Applied Sciences Ulm, Germany, Michael.munz@thu.de moves away from the wall. Although prolongating the fall Thomas Engleder: University of Applied Sciences Ulm, Germany Open Access. © 2019 Michael Munz et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution-NonCommercial- NoDerivatives 4.0 License. M. Munz et al., Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing for Injury Prevention based Se nso r D ata — 184 on Inertial length, dynamic belaying reduces the maximum deceleration 2.2 Execution of Trials force. Hard belaying introduces additional force into the Several trials were executed in order to prove the basic system, resulting in a strong impact to the climber towards feasibility of metrological acquisition of the required the wall, which often leads to injuries of the climber. quantities by means of IMUs. Forces are evaluated by means Therefore, the dynamic of the belay technique is important to of force measuring sensors. However, this should later be prevent injuries [5]. completely replaced by IMU measurements. The quantities to be acquired by the system are: In total, 16 trials with different conditions were carried out in  Forces on the climber a climbing gym during toprope and lead falls (see table 1). In  Fall duration, fall distance toprope, some of the falls were deliberately soft or very  Accelerations on the legs of the climber during impact dynamic, some passive or even with hard belay. The fall  Trajectory of the climber during fall (rotation, etc.) lengths in toprope situation were about 2 m each. This  Movement of the belayer (upwards, downwards, etc.) corresponds to the amount of slack rope deliberately issued by the belayer. In the lead, the last quickdraw was exceeded Sensors can determine these quantities metrologically. From by 0.7 m to 1.5 m, resulting in fall lengths of 1.5 m to 3 m. the measured data, quality criteria (scores) for the fall and the IMU sensors were attached to the body's centre of gravity belay process are to be calculated and evaluated in a next (COG) of both climber and belayer and to the climber's step. These are among others: ankles. The attachment of the sensor centrally on the back of  Strength of the impact, dynamics of belaying the waist belt is convenient to map the attack of the  Adequacy of the amount of slack rope gravitational force at the centre of mass. The acceleration  Jumping behaviour of the climber (too far away from the sensors on the ankles should provide additional information wall, rotation, etc.) about the impact energy when contacting the wall. In  Impact on the wall and force reduction addition, the load cell was inserted into the safety chain at the  Stance position of the belayer climber’s central belay loop at the harness. In this paper, we  Reaction of the belayer (passive / active, etc.) only present climber’s data. The scores should provide an objective measure of the In a climbing gym, DIN 2015 / 2015-07-00 defines the quality of the fall course, broken down into climbers and maximum distance between two quickdraws with height > belayer aspects. 8m above ground to 1.50 m. The trial conditions in table 1 therefore represent the maximum expected fall height in climbing without unnecessary slack rope in the system. 2 Material and Methods Table 1: List of trial conditions 2.1 Measurement principle No Type Free rope Slack Height above length [m] rope last quickdraw Measurements include force, acceleration, orientation and [m] [m] short-term position, and their temporal change. To detect 1 Dynamic 15,0 2,0 - forces usually load cells are used, which are inserted into the 2 belay 15,0 2,0 - safety chain. However, this introduces an additional risk 3 Hard belay 15,0 2,0 - (breakage, wrong integration, etc.) and affects belayer and 4 15,0 2,0 - climber in their normal behaviour. Therefore, in this work we 5 Wide jump 15,0 2,0 - 6 15,0 2,0 - show that forces can be reconstructed from the IMU data. To 7 “Normal” 12,0 2,0 1,0 validate this method, a load cell with a measuring range of ± 8 conditions 12,0 - 1,0 10 kN was used. 9 12,0 - 1,0 To measure the movement and accelerations, IMU 10 Leading 6,5 - 0,7 sensors (MPU-9150, InvenSense) were attached to climber climb with 11 8,0 - 0,7 and belayer. Each IMU consists of three-axis acceleration, different 12 9,0 - 1,5 gyro and magnetic field sensors. All data is digitally recorded conditions 13 10,0 - 0,7 14 11,0 - 1,5 by a microcontroller and transmitted wirelessly via Bluetooth 15 11,0 - 0,7 to a computer, allowing synchronous data acquisition of 16 12,0 - 1,5 several IMUs. The measurement rate was 100 Hz. M. Munz et al., Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing for Injury Prevention S ensor Data — 185 based on Inertial the integration is realized using a numerical integration method. The initial impulse of the fall is fit using linear regression, which corresponds to an unrestrained fall. The greatest deviation of the impulse from the regression line is at impact into the rope, which leads to the braking timestamp. Subsequently, the beginning and end of the fall are detected. For this purpose, a resting position is detected for the end of the fall, when only gravitational acceleration is measured for a certain period of time (after the decay phase). The start of the free fall is detected by comparing to a low threshold (additional acceleration due to the rope friction and the redundant belay needed for the trials). By this, the duration of the free fall is defined. The fall length can be calculated by double-integrating the acceleration. This is applicable because of the short integration time and the low acceleration noise. With the known body weight m of the climber, the force at the central belay loop of the climber can be estimated from the resulting acceleration using . The timestamps of the maxima during the deceleration Figure 1: Resulting acceleration (yellow) and segmented process (impact force) in both and measured force are phases / events during fall: jump start (blue), jump end determined. The time offset and the difference between the (magenta), max. acceleration (green) measured at the COG of the climber. Top: hard belay, bottom: dynamic maximum values serve as a measure of the correspondence belay. between the estimated value and the real force measurement . 2.3 Data processing From the data of the inertial sensors it is possible to deduce the orientation of the sensor at any time. For this, a gradient 3 Results and discussion decent algorithm [6] for fusing acceleration and gyro The evaluation of the force measurements shows a temporal measurements of the IMU sensors is used. In addition to the deviation of 0.01 s (median) with a range of -0.04 to +0.03 s accelerations, this also allows the body movements and body (1st or 3rd quartile) over all measurements. The duration of position to be determined, which provide essential information for the later scores. the impact peak according to [7] is between 0.1 s and 0.2 s. Therefore, the results are within the time window of the The evaluation of the measurement results focuses primarily on the measurement data of the force sensor and impact peak, and so the maximum impact was always detected correctly. the acceleration values at the COG. Supplementary information is obtained by the determined acceleration values For the deviation of the estimated force from the at the ankles. The coordinate axes of the acceleration sensors measured force , the following values were obtained for all span a Cartesian coordinate system. The resulting absolute experiments: Median: -112 N, range: -512 N to -102 N (1st acceleration can be calculated based on the accelerations and 3rd quartile, respectively). In addition to the force peaks, along the x-, y- and z-axis using = the course of the estimated and measured force was also examined. For this purpose, the course from the beginning of Based on , the fall can be segmented into different the deceleration process to the capture impact was subjected phases, as shown in figure 1. Start and end of the fall, time to a congruence analysis. Figure 2 shows results of the and strength of the maximum impact into the rope are individual trials as box plots. All values are normalized to the automatically calculated by a segmentation algorithm. For maximum observed force of 2050 N. this purpose, various filters and pattern recognition The difference during the phase of the free fall is due to techniques are applied to the measurement data. the rope friction in the carabiners of the quickdraws, leading First, the data of the acceleration sensor is low pass to greater differences in trials with longer falls. filtered for noise reduction. Afterwards, the impulse is Analysing the acceleration data in more detail, additional estimated from the acceleration data using , where assessment can be conducted. The strength of the impact M. Munz et al., Intelligent Assistant System for the Automatic Assessment of Fall Processes in Sports Climbing for Injury Prevention based on Inertial Sensor Data — 186 Figure 3: Example of a fall with asynchronous impact of feet Figure 2: Boxplot of the difference between measured force F after the main deceleration peak of the rope impact. and FB for all 16 experiments. The deviation between F and (green). is the amount of slack in a fall. This aspect must be assessed FB normalized to 2050 N is indicated on the y axis. in relation to the climbing height. Sensory extensions are planned, which will be evaluated in subsequent work, peaks allows the differentiation between soft and hard allowing an evaluation of a possible fall situation at any time. belaying. Unnecessarily strong initial take-off of the climber Further work has been done in developing algorithms for can also be detected using acceleration data and the direction scoring relevant quality criteria for safer and climbers on the of acceleration. By measuring the accelerations on the ankles, presented features extracted from acceleration data. The the delay between impact of both feet on the wall can be output of this scoring algorithm can then be evaluated as part detected. This serves as an indication of an unstable posture of a series of measurements with independent climbing and thus a poor control of the fall. Figure 3 shows a coaches. corresponding example: after the impact and breaking start in the rope, the feet touch the wall asynchronously (two peaks). Author Statement In addition, the strength of the impact of feet on the wall can Research funding: The authors state no funding involved. be determined. The shape of this peak can also serve as a Conflict of interest: Authors state no conflict of interest. feature for assessment of the impact attenuation by the climber’s feet. References 4 Conclusion and Outlook [1] Hummel, C. & Hellberg, F. „Unfälle an künstlichen Kletterhallen: Kampf dem Eisberg!“. DAV Panorama: Magazin des Deutschen Alpenvereins 6/2014, 62–65. In this work, a prototype of a fall and safety instructor for the [2] Schöffl, V.R., Hoffmann, G. & Küpper, T. (2013): Acute injury quantitative assessment of forces and movements in sport risk and severity in indoor climbing-a prospective analysis of climbing falls is presented, providing objective feedback to 515,337 indoor climbing wall visits in 5 years. – Wilderness & the athletes or coach. To the best of the authors’ knowledge, environmental medicine 24, 3, 187–194. [3] Radelzhofer, P. (2014500): Bergunfallstatistik 2012-2013. this is the very first system that can perform such an München. assessment in sports climbing for injury prevention. [4] Tonoli (2011): Fall identification in rock climbing using Validation measurements show that the impact force at wearable device. In: Journal of Animal Science (X), S. 1–11. DOI: the climber's rope point can be estimated using IMU 10.1177/ToBeAssigned. [5] Schöffl, Volker; Morrison, Audry; Schöffl, Isabelle; Küpper, measurement data, allowing to quantify the softness of the Thomas (2012): The epidemiology of injury in mountaineering, belaying technique, as one important factor for injury rock and ice climbing. In: Medicine and sport science 58, S. 17– prevention. With a first analytical evaluation of the IMU 43. DOI: 10.1159/000338575. sensor data, we showed that it is possible to differentiate [6] Madgwick, S. O., Harrison, A. J., & Vaidyanathan, R. (2011): various processes and effects, allowing assessment of both “Estimation of IMU and MARG orientation using a gradient descent algorithm”. In Rehabilitation Robotics (ICORR), 2011 belay and fall quality. IEEE International Conference on (pp. 1-7). IEEE. A frequently discussed, very important parameter that [7] Fimml, W. & Larcher, M. (2000): „Energie ist Kraft mal Weg: cannot yet be measured and evaluated with the current Sicherungstheoretische Grundlagen, Teil 2“. bergundsteigen: system Zeitschrift für Risikomanagement im Bergsport 4/00, 14–20.

Journal

Current Directions in Biomedical Engineering – de Gruyter

Published: Sep 1, 2019

Keywords: Accident prevention; sports climbing; fall analysis; wireless body sensor networks; inertial measurement units

There are no references for this article.