An experimental approach into the quantification of steering and balance behaviour of bicyclists

Research output: ThesisDissertation (TU Delft)

10 Downloads (Pure)


The aim of this thesis is to derive bicycle rider control models, based on experimental data, that mimic the rider in his balance control task at various forward speeds. These rider control models can help to understand cyclists falls, improve training techniques, assess the handling properties of new bicycle designs and create active balance control systems (e.g. steer assist). This thesis consists of 9 chapters; Chapter 1 introduces relevant background theory and identifies the research gap.

Chapter 2 presents some effects of crosswind on the lateral dynamics of a bicycle and on rider control. The chapter gives an insight on how rider control modelling can be used to assess crosswind related falls. Simulations indicated that crosswind has a considerable effect on the stability and control of the bicycle. Increasing wind speed can make an uncontrolled bicycle resonate for all forward speeds. The rider control effort increases considerably and a constant steer torque is required to keep the bicycle at a straight heading.

Chapter 3 investigates the dynamic response of the bicycle rider’s body during vertical, fore-and-aft and lateral perturbations in order to understand how riders are using postural control to restrain excessive movements and prevent falling off the seat. The analysis is presented by means of apparent mass (APMS) and seat-to-sternum transmissibility (STST) functions in the frequency domain. Measured forces at saddle, steer and pedals revealed that for each individual motion the rider applied forces in all three directions. Heave and surge motion interacted with each other and had similar responses. Sway showed totally different responses and weak interaction with the other two motions. Resonant frequencies were considerably higher in the vertical direction as compared to the longitudinal direction. Lateral measurements showed no resonance, and trunk postural control was evident in the APMS. The results of this chapter can be used to identify the parameters of biodynamic lumped human-machine models. Such models can support the development of more comfortable and safe bicycle designs and suspension systems.

Chapter 4 presents the design and implementation of an instrumented steer-by-wire bicycle (SBW) that was designed and built at TU Delft bicycle laboratory. The
SBW was used as a versatile experimental platform to capture the rider’s responses with (haptics on) and without steering torque feedback (haptics off) during lateral perturbation experiments. Simulations and testing of the steer-by-wire system indicated good tracking performance between 0-2.5 Hz and almost identical steer stiffness with the Carvallo Whipple bicycle model in a frequency range of 0-3 Hz and in a forward speed range of 0-10 m/s. The bicycle served its purpose successfully, the responses of the rider’s control actions with lateral perturbations were captured by means of impulse response functions (IRFs) in chapter 5. Results failed to indicate any statistically significant difference between the two steering configurations (haptics on/ off).

Chapter 6 presents and validates a parametric rider control model using data presented in Chapter 5 and uses this model to further assess the effect of haptic
feedback in the balance task of bicycling. Bicycle and rider mechanics have been modelled using the Carvallo Whipple bicycle model extended with rider inertia. A
balancing and heading controller was added, capturing visual, vestibular and proprioceptive sensory information using feedback of roll angle, roll angle rate, heading angle, heading angle rate, steering angle and steering torque, taking into account muscular activation dynamics. Non-parametric and parametric model responses failed to indicate any statistically significant difference between the haptics on/off configurations. However, further analysing the haptic off configuration it became apparent that the rider still receives relevant torque feedback due to the inertia of the handlebars. The reduced feedback was proven to be adequate for the rider to control the bicycle without any major steering discrepancies. To further evaluate the effect of torque feedback in simulations we disconnected the handlebar torque feedback loop of the parametric rider model. In addition, we also disconnected the handlebar position and velocity feedback. Results showed that handlebar torque feedback is significantly important during the riding process. This knowledge might be crucial for the development of new safety systems that could further optimize bicycle handling and assist the rider’s steer control actions in critical situations preventing falls.

Chapter 7 outlines the design and hardware selection for a bicycle simulator. The design requirements together with a detailed description of the hardware selection and testing are presented. The simulator was designed to explore human control behaviour in a safe environment. Preliminary tests showed that all subjects can balance and manoeuvre the bicycle when a simplified bicycle model is used to generate haptic feedback and project the dynamics in the virtual environment. Visual roll of the horizon turned out to be an effective tool for creating the illusion of physical roll but motion sickness was reported.

This thesis ends with the discussion and conclusion Chapters 8, 9 highlighting the developed experimental facilities and the main findings of the research. The chapters herein investigate the effects of external perturbations on bicycle stability and human control using numerical modelling and experimental bicycles capable of measuring kinematics and rider applied forces. This interdisciplinary approach delves into the foundations of human control modelling from both a biomechanical and biomechatronics engineering perspective in an effort to improve cycling safety and reduce falls.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Delft University of Technology
  • Happee, R., Supervisor
  • Schwab, A.L., Advisor
Award date15 Sep 2020
Publication statusPublished - 2020


  • bicycle dynamics
  • bicycle control
  • human biomechanics

Fingerprint Dive into the research topics of 'An experimental approach into the quantification of steering and balance behaviour of bicyclists'. Together they form a unique fingerprint.

Cite this