ArticlePDF Available

Measuring and Comparing Descend in Elite Race Cycling with a Perspective on Real-Time Feedback for Improving Individual Performance

Authors:

Abstract and Figures

Descend technique and performance vary among elite racing cyclists and it is not clear what slower riders should do to improve their performance. An observation study was performed of the descending technique of members of a World Tour cycling team and the technique of each member was compared with the fastest descender amongst them. The obtained data gives us guidelines for rider specific feedback in order to improve his performance. The bicycles were equipped with a system that could measure: velocity, cadence, pedal power, position, steer angle, 3D orientation, rotational speeds and linear accelerations of the rear frame and brake force front and rear. From our observation study, the brake point and apex position turned out to be distinctive indicators of a fast cornering technique in a descent for a tight, hairpin corner. These two indicators can be used as feedback for a slower rider to improve his descend performance.
Content may be subject to copyright.
Proceedings 2018, 2, 262; doi:10.3390/proceedings2060262 www.mdpi.com/journal/proceedings
Proceedings
Measuring and Comparing Descend in Elite Race
Cycling with a Perspective on Real-Time Feedback
for Improving Individual Performance
M. M. Reijne *, D. J. J. Bregman and A. L. Schwab
Department of BioMechanical Engineering, Delft University of Technology, 2628 CD Delft, The Netherlands;
D.J.J.Bregman@tudelft.nl (D.J.J.B.); A.L.Schwab@tudelft.nl (A.L.S.)
* Correspondence: m.m.reijne@tudelft.nl; Tel.: +31-152-786-400
Presented at the 12th conference of the International Sports Engineering Association, Brisbane,
Queensland, Australia, 26–28 March 2018.
Published: 14 February 2018
Abstract: Descend technique and performance vary among elite racing cyclists and it is not clear
what slower riders should do to improve their performance. An observation study was performed
of the descending technique of members of a World Tour cycling team and the technique of each
member was compared with the fastest descender amongst them. The obtained data gives us
guidelines for rider specific feedback in order to improve his performance. The bicycles were
equipped with a system that could measure: velocity, cadence, pedal power, position, steer angle,
3D orientation, rotational speeds and linear accelerations of the rear frame and brake force front and
rear. From our observation study, the brake point and apex position turned out to be distinctive
indicators of a fast cornering technique in a descent for a tight, hairpin corner. These two indicators
can be used as feedback for a slower rider to improve his descend performance.
Keywords: cycling; descending; cornering; cyclist technique; bicycle control; training feedback
1. Introduction
Some professional road cyclists descend faster than other riders. These differences can clearly be
seen in races. Unfortunately, coaches are often in doubt on the specific advice they can give to slower
descenders such that they can improve their performance.
Literature on descend and cornering technique for cyclists is non-existent, but on social media a
lot of advice can be found. Some advice examples are: looking ahead to unconsciously ride the
optimal trajectory or pushing hard on the outer crank and inside handlebar to maximize tire grip [1].
Common practice by coaches are to have a slow descender follow and mimic a fast descender.
A better solution would be to give real-time feedback to the rider based on the data of an
observation study where the differences in accelerations, decelerations, position on the road and
steering technique are measured between fast and slow riders. Large differences highlight specific
actions the slower rider can improve. This paper presents the results of a first observation study. The
work is a continuation of the seminal work by Lommers [2].
This paper is organized as follows. First, the experiment and measurement system are described
in the materials and methods section. Next the results, being the time differences between the rider,
differences in trajectory through a corner and differences in brake force and braking point are
presented. The paper ends with a discussion and conclusions section.
Proceedings 2018, 2, 262 2 of 6
2. Materials and Methods
The aim of the project is to provide professional cyclists with real-time feedback on how they
can improve their descent time. Data of the descend technique of fast and slow riders is required in
order to identify performance indicators. Therefore, an observation study was performed with eight
professional cyclists from Team Sunweb. All participants signed an informed consent. For this study,
the bicycle of each rider was equipped with the measurement system as shown in Figure 1. Each rider
rode a 1.85 km descent, as shown in Figure 2, six times on his instrumented bicycle. They were
instructed before each trial to cycle and descend as fast as possible. The following safety procedures
were taken to ensure the safety of the riders. First, all riders performed the descent on their own,
familiar bike. Each rider was familiar with the decent. Furthermore, before each trial the track was
cleaned from debris. The experiment was performed in dry conditions only. The riders were told to
cycle as fast, but safe as possible. Lastly, the riders were directly followed by a car with a person that
could provide medical assistance if necessary.
Figure 1. Measurement system installed on a bicycle. (A) GoPro camera and mount to record rider
posture; (B) GNSS antenna to measure rider’s position on the road; (C) optical encoder to measure
steering angle; (D) IMU to measure roll angle and data acquisition system; (E) strain gauges on brake
shoe to measure brake force; (F) velocity sensor and power meter.
2.1. Measurement System
The bicycles were equipped with a measurement system that could measure: velocity, cadence
and pedal power, position, steer angle, 3D orientation, rotational speeds and linear accelerations of
the rear frame, brake force front and rear, and rider posture. The position of the rider on the road is
measured with a Global Navigation Satellite System (GNSS) receiver. To increase accuracy, a base
station was used and the GNSS antenna was mounted high above and behind the rider (see Figure 1).
Even so, occlusions remained a problem on certain areas of the descent due to the environment (e.g.,
trees or proximity of a high cliff). The front and rear brake forces of the rider are measured using
custom brake shoes on which strain gauges are installed. The lean angle of the bicycle is measured
using an Inertial Measurement Unit (IMU) positioned at the bottle cage holder (see Figure 1). An
optical encoder measures the steering angle. The back-wheel velocity and power produced by the
A
E
E
C
B
D
F
Proceedings 2018, 2, 262 3 of 6
athlete are measured by a velocity sensor and power meter standardly equipped on the team bicycles.
Finally, a GoPro camera was mounted above and behind the rider to record his posture during the
descent. The power meter and velocity sensor record at 1 Hz, the GNSS receiver at 5 Hz, the GoPro
at 90 Hz and all other sensors at 100 Hz. All data is locally stored on a removable flash drive. The
measurement system is interchangeable between bicycles within 10 min.
Figure 2. Location of the experiment is a 1.85 km part of road of the L281 near Vossenack, Germany.
The average gradient of the descent is 5 degrees. Start location is at the top left. The descent is divided
into eight segments to compare rider technique for different elements (e.g., straight parts, left hand
corners, right hand corners).
2.2. Data Processing
The control actions of the riders are compared at the same travelled track centerline distance as
time and position on the road vary between riders and trials. Therefore, the position of the bicycle is
projected on the track centerline. The track centerline is reconstructed from GNSS coordinates of the
track boundaries, which is measured before the experiment, to a curvilinear coordinate system using
the method described by Perantoni and Limebeer [3,4].
3. Results
First the time differences between the riders are compared. Figure 3 shows the time differences
between each rider, for his fastest trial, as a function of the travelled track distance. Rider 5 had the
fastest trial time and is therefore used as a baseline. If the gradient in Figure 3 is negative, this implies
that a rider is slower than the fastest rider, whereas a positive gradient is the opposite. Consequently,
the gradient coefficient shows how much time is won or lost. The vertical distribution of the lines
also represents the ranking of descent time of the riders. It can be seen that rider 5 gains the most
time on the other riders on the straights (segment 1 and segment 5), but loses time in the turns
(segment 2, 6 and 8). The turns in the track lead to many changes in the ranking of the riders.
Proceedings 2018, 2, 262 4 of 6
Figure 3. The time difference between each rider, for his fastest trial, as function of the travelled track
distance. The trial of rider 5 is taken as the baseline to calculate the time differences between the riders.
The track distance is divided into six segments. Segment 2, 4 and 6 are sharp corners.
Segment 6 is a corner with the smallest radius of the track. This corner is therefore evaluated in
more detail. Riders 1, 6 and 8 had the fastest split times in this corner, and rider 2, 3 and 7 the slowest
split times. Figure 4 shows the trajectories of all trials of rider 5 and rider 6 in segment 6. Rider 6 was
consistently faster than rider 5 in this corner. The trials of rider 5 and 6 have little trajectory variation.
Rider 5 consistently takes a wider trajectory through the corner for all trials compared to rider 6.
Entry and exit position are similar for both riders. Also, for both riders, the trajectory of the fastest
trial is closest to the right boundary road.
Figure 4. The trajectories of all trials of rider 5 and rider 6 for segment 6 (turn 3). This is the sharpest
right-handed turn of the descent. The trajectory of the fastest trial of each rider is bold and highlighted
with a marker. The track boundaries for the right lane of the road are plotted in black. The dotted line
indicates the centerline of the road.
Proceedings 2018, 2, 262 5 of 6
Figure 5 shows the normalized brake force of the fastest trial of each rider superimposed on the
track centerline for segment 6. The brake force of rider 5 is missing as the brake sensors could not be
installed on his bicycle and the brake force of rider 3 is missing as the calibration was forgotten before
his fastest trial. It can be seen that riders 2 and 7 brake harder and earlier than the other riders. Rider
1 and 6 brake the latest.
Figure 5. The normalized brake force applied by the rider in segment 6 during his fastest trial
superimposed on the centerline of the descent. Riders 1, 6 and 8 have the fastest split times for this
segment. Riders 2, 3 and 7 had the slowest split times. Note that the brake data of rider 5 and rider 3
is not available.
4. Discussion
The measurement system turned out to be very robust during the experiment. All sensors
maintained working throughout all trials, even in extreme conditions. However, the GNSS sensors
encountered occlusions on several parts of the track, sometimes resulting up to a gap of five seconds
without position fix. This resulted in loss of information for turn 2 (segment 4) which would have
been an interesting part of the track to compare rider technique as there are smaller time differences
between riders for this corner. The data of the IMU sensors might be used in the future, to determine
the rider’s position on the road during short occlusion periods.
In addition, the ANT+ protocol proved to be unreliable to transfer data from the power meter
and velocity sensor to the flash drive. When the power meter and velocity sensor went on standby
(between trials of the same rider), the connection was not maintained but cut-off. The power meter
and velocity data was retrieved from the power meter head unit and synchronized manually with
the rest of the data.
The brake point and apex position are promising indicators of cornering performance. However,
the importance of these two indicators remain corner dependent. For example, braking becomes less
important when the approach velocity for a corner is lower and the fastest trajectory through a corner
also depends on the previous and upcoming corners.
5. Conclusions
From our observation study, it can already be seen that there are two distinctive indicators of a
fast cornering technique in a descent for a tight, right handed, hairpin corner. The fastest riders
Proceedings 2018, 2, 262 6 of 6
through this corner brake 10–20 m later. The fastest rider also steers in closer to the corner’s inside
road boundary for similar corner entry and exit position on the road. Therefore, feedback on the brake
point and apex position can be used to improve individual rider descent performance. Future work
will be directed towards developing a system for real-time feedback.
Acknowledgments: This project was funded by “Nationale Sportinnovator prijs 2017”. We would also like to
thank Team Sunweb for their support, collaboration and input and Niels Lommers for starting the project and
developing the first prototype.
Conflicts of Interest: The authors declare no conflict of interest.
References
1. Global Cycling Network. Available online: https://www.globalcyclingnetwork.com/video/how-to-descend-
like-a-pro (accessed on 6 October 2017).
2. Lommers, C.R. Descending: Measuring and Comparing Descending Technique and Performance in
Professional Road Cycling. Master’s Thesis, Delft University of Technology, Delft, The Netherlands,
November 2015.
3. Perantoni, G.; Limebeer, D.J. Optimal control for a formula one car with variable parameters. Veh. Syst. Dyn.
2014, 52, 653–678.
4. Perantoni, G.; Limebeer, D.J. Optimal control of a formula one car on a three-dimensional track—Part 1:
Track modeling and identification. J. Dyn. Syst. Meas. Control 2015, 137, doi:10.1115/1.4028253.
© 2018 by the authors; Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
... Studies of vehicle drivers show that the tangent point on the inside of a bend, beyond which one cannot see the full breadth of the road, is an observational focus [8]. In cycling, faster riders over a set course brake later at corners and use the full width of the road [9]. ...
... In the remaining sections we analyse that model, specifically with the objective to find a generic estimate of t * to minimise centripetal acceleration along routes. This is a safety-first approach to using pedalling and braking at bends, rather than minimising total time by braking as late as possible [9]. One potential issue is that for a region of large curvature, the rider may be unable to see beyond the bend, limiting t * . ...
Article
Full-text available
We present a mathematical model of road cycling on arbitrary routes using the Frenet–Serret frame. The route is embedded in the coupled governing equations. We describe the mathematical model and numerical implementation. The dynamics are governed by a balance of forces of gravity, drag, and friction, along with pedalling or braking. We analyse steady-state speed and power against gradient and curvature. The centripetal acceleration is used as a control to determine transitions between pedalling and braking. In our model, the rider looks ahead at the curvature of the road by a distance dependent on the current speed. We determine such a distance (1–3 s at current speed) for safe riding and compare with the mean power. The results are based on a number of routes including flat and downhill, with variations in maximum curvature, and differing number of bends. We find the braking required to minimise centripetal acceleration occurs before the point of maximum curvature, thereby allowing acceleration by pedalling out of a bend.
... The group of Professor A. Schwab at the Delft University of Technology has done scientific research on the topic [6,7]. However, their research has been focused on the bike-rider interaction from a 'vehicle dynamics' perspective, rather than a 'sport engineering' perspective. ...
Article
Performance in downhill road cycling is understudied. Tools that can be used to assess the cyclists’ cornering strategies ecologically and objectively are missing. A new methodology based on motion capture and mathematical modelling is presented here. A drone was used to capture the trajectory of the centre of mass of a cyclist, who was asked to complete 10 times a ~220-m-long downhill course. The motion capture and ‘optimal’ trajectories were compared in terms of displacement, speed, and heading. In each trial, the apex, the turn-in and the braking points were detected. Whilst the ‘optimal’ trajectory suggested an ‘early’ apex strategy was best, the cyclist in this study completed the corners with a ‘late’ apex strategy. This study presents a methodology that can be used to objectively assess cornering strategies in road cycling. Discrepancies between actual and ‘optimal’ trajectories are also discussed. This study brings to light concepts such as: ‘early’ or ‘late’ apex, braking and turning points, which are discussed within the context of road cycling downhill performance.
... Few studies have investigated racing trajectories in cycling, 40 whilst trajectories are well documented in racing cars, 41 where two cornering strategies can be detected: (I) steady velocity and (II) high entry and exit velocities. A cycling analogy can be made here. ...
Article
A mathematical model of a bike-rider's longitudinal and lateral dynamics was used to study the influence of road conditions (tyre-road friction coefficient) on cycling individual time trial (ITT) performance and pacing strategy. A dynamic optimisation approach was used on different simulated 40-km-ITT courses, where environmental variables (i.e. slope and wind), the presence of corners and the tyre-road friction coefficient were varied. The objective of the optimisation was the performance time. Maximal velocity was constrained by road geometry and the tyre-road friction coefficient. The maximal deliverable power output was constrained accordingly to the critical power model. The simulation results suggest that when technical sections constitute 25% of the entire course, road conditions can meaningfully affect the final performance time and peak power required, but not the pacing strategy. In fact, the time lost in slow technical sections cannot be regained during fast straight sections, even if technical sections are used to restore anaerobic energy stores. However, more experimental research is needed to test the applicability of these predictions.
Article
Purpose The purpose of this research was to investigate the beliefs, attitudes, and experiences of stakeholders in youth triathlon regarding the important motor subskills that are required to be successful at the elite level of triathlon competition. Method Twenty-five participants were recruited from five stakeholder groups in triathlon and interviewed via video conference. A constructionist and relativist approach to thematic analysis was used to identify three first order themes and several second order themes. Results The first, first order theme was ‘Continuous motor skills' which consisted of the invariant features of triathlon's continuous motor skills and the parameterization of continuous motor skills. The second, first order theme was ‘Discrete Motor Skills' and consisted of discrete motor skills involved with cornering and change of direction in each discipline and transition phases in triathlon. The final first order theme was ‘Adaptability to continuous and discrete motor skills'. Conclusion This research provides a novel and more broad understanding of the beliefs, attitudes, and experiences of stakeholders in triathlon regarding important motor skills that are required to succeed at the elite level of the sport. This novel and broad understanding of important triathlon motor skills has theoretical implications for evaluating triathlon performance with skill acquisition as a primary focus. Additionally, this research is practically important for coaches, administrators, and athletic performance staff who design training programs and pathways for young, developing triathletes.
Article
The identification of three-dimensional (3D) race track models from noisy measured GPS data is treated as a problem in the differential geometry of curves and surfaces. Curvilinear coordinates are adopted to facilitate the use of the track model in the solution of vehicular optimal control problems. Our proposal is to model race tracks using a generalized Frenet-Serret apparatus, so that the track is specified in terms of three displacement-dependent curvatures and two edge variables. The optimal smoothing of the curvature and edge variables is achieved using numerical optimal control techniques. Track closure is enforced through the boundary conditions associated with the optimal control problem. The Barcelona formula one track is used as an illustrative example.
Article
The minimum-lap-time optimal control problem for a Formula One race car is solved using direct transcription and nonlinear programming. Features of this work include significantly reduced full-lap solution times and the simultaneous optimisation of the driven line, the driver controls and multiple car set-up parameters. It is shown that significant reductions in the driven lap time can be obtained from track-specific set-up parameter optimisation. Reduced computing times are achieved using a combination of a track description based on curvilinear coordinates, analytical derivatives and model non-dimensionalisation. The curvature of the track centre line is found by solving an auxiliary optimal control problem that negates the difficulties associated with integration drift and trajectory closure.
Available online: https://www.globalcyclingnetwork.com/video/how-to-descendlike-a-pro
  • Cycling Global
  • Network
Global Cycling Network. Available online: https://www.globalcyclingnetwork.com/video/how-to-descendlike-a-pro (accessed on 6 October 2017).
Descending: Measuring and Comparing Descending Technique and Performance in Professional Road Cycling
  • C R Lommers
Lommers, C.R. Descending: Measuring and Comparing Descending Technique and Performance in Professional Road Cycling. Master's Thesis, Delft University of Technology, Delft, The Netherlands, November 2015.