Proposal for ASME DSCC/MOVIC 2012 Invited Session on

Single Track Vehicle Dynamics and Control: the Human Rider

Session Organizers

Stephen M. Cain, Andrew Dressel, Mont Hubbard, Jason K. Moore, Arend L. Schwab 

We propose an invited session for the 2012 ASME Dynamic Systems and Control Conference and 2012 Motion and Vibration Conference on the topic of single track vehicle dynamics and control: the human rider. This session is one of two proposed sessions on single track vehicles and would include research focused on understanding the needs and behavior of the human rider and includes both experimental and analytical work. The session organizers and many of the authors of the included papers participated in Bicycle and Motorcycle Dynamics 2010 (BMD2010), an international three-day symposium on the dynamics and control of single track vehicles held at the Delft University of Technology. This proposed session would have a similar goal to BMD2010, in that it would provide an opportunity for the members of the single track vehicle research community to interact in way not possible without a focused session or conference. We have included short abstracts of seven papers that will be submitted as possible papers for the session.


Stephen M. Cain, University of Michigan, Ann Arbor

Andrew Dressel, University of Wisconsin-Milwaukee

Mont Hubbard, University of California, Davis

Jason K. Moore, University of California, Davis

Arend Schwab, Delft University of Technology

Session Abstract

Single track vehicles such as bicycles and motorcycles leave a single track on the ground as they roll forward, unlike other types of vehicles such as cars. The dynamics of single track vehicles are unique; typically the steer and roll/lean of a vehicle are coupled and the stability of the vehicle is velocity dependent. Control of single track vehicles is not trivial, and some single track vehicles, especially motorcycles, exhibit unstable modes that can be very difficult to suppress. This session will primarily focus on research about the lateral dynamics of motorcycles and bicycles, and aims to contribute to the understanding of manually controlled single track vehicle systems. Specifically, these studies investigate: methods to quantify rider skill, rider control models, a method to help a human rider stabilize a motorcycle, methods to assess stability of human/bicycle systems, rider neuromuscular dynamics, and human power requirements. We believe that the papers included in this session provide an excellent overview of the current research in single track vehicle dynamics directed towards understanding the human rider.

Papers in the Session

We have a total of seven papers in this proposed invited session.  They are:

  1. S. M. Cain, D. A. Ulrich, and N. C. Perkins, “Using measured bicycle kinematics to quantify increased skill as a rider learns to ride a bicycle”
  2. A.L. Schwab, P.D.L. de Lange, and Jason K. Moore, “Rider optimal control identification in bicycling”
  3. Shintaroh Murakami, Hidekazu Nishimura and Shaopeng Zhu, “Stabilization control of a motorcycle during braking”
  4. Adrian Cooke, Vera Bulsink, Rosemary Dubbeldam, Bart Koopman, Wim Poelman, Marc Beusenberg and Maarten Bonnema, “Methods to assess the stability of a bicycle/rider system”
  5. Jingang Yi, Damoon Soudbakhsh, Yizhai Zhang, and Yang Zhang, “Why some Parkinson’s disease patients cannot stand or walk but can ride a bicycle - A control system-based analysis”
  6. M. Massaro and D.J. Cole, “Steering neuromuscular dynamics: motorcycle riders vs car drivers”
  7. Junghsen Lieh, “Closed form and numerical solutions of traction in cycling”


Using measured bicycle kinematics to quantify increased skill as a rider learns to ride a bicycle

S. M. Cain, D. A. Ulrich, and N. C. Perkins

University of Michigan, Ann Arbor, MI, USA

Currently, it is difficult to determine when a novice bicycle rider is ready to ride without training wheels or external assistance.  In this study, we quantify the changes that occur as a rider learns to ride a traditional bicycle by employing three synchronized wireless inertial measurement units (IMUs) to measure bicycle kinematics. Our study considers 10 subjects with disabilities who learned to ride bicycles during a specialized bicycle training camp.  The 10 subjects had a wide range of disabilities including Down syndrome (n=3), autism spectrum disorder (n=5), cerebral palsy (n=1), and attention deficit hyperactivity disorder (n=1). The training camp, run by the organization Lose the Training Wheels (, utilizes adapted bicycles with crowned rollers in place of a rear wheel.  Unlike training wheels, the crowned rollers allow the bicycles to roll/lean similar to traditional bicycles.  As riders show improvement, the adapted bicycles function similar to traditional bicycles by increasing the gearing and using rollers with greater crown.  Subjects attended the camp five consecutive days for a 75 minute training session each day.  Out of 10 subjects, 6 were successful in riding a traditional bicycle without assistance by the end of the training camp. We used the IMUs to measure bicycle kinematics during each training session for each rider. Our goal was to disrupt the training session as little as possible and therefore no special instruction or explanations were given to the riders.  The IMU measurements resolved the bicycle frame roll rate, steer rate, and bicycle speed.  We calculated the average speed, standard deviations of steer and roll angular velocities and normalized cross-correlation between steer and roll angular velocities.  The peak value of the cross-correlation between steer and roll angular velocities increased significantly with time (t = 5.944, p < 0.001) and was significantly greater for riders who ultimately succeeded in riding a traditional bike without assistance (t = 3.567, p = 0.009). This finding suggests that rider learning is quantified by increased correlation between bicycle steer rate and roll rate.  In essence, learning to steer in the direction of lean is an essential skill in learning to ride a bicycle. Average speed increased with time (t = 4.591, p = 0.001) and the standard deviation of the roll rate increased with time (t = 3.864, p = 0.004), likely due to the increased gearing and more crowned rollers used as a rider progresses through the camp.  The standard deviation of the steer rate also increased with time (t = 2.690, p = 0.025), suggesting that initially fearful riders learn to relax their arms and use the handlebars to control and balance the bicycles.

Rider optimal control identification in bicycling

A.L. Schwab1, P.D.L. de Lange1, and Jason K. Moore2

1Delft University of Technology, Delft, The Netherlands

2University of California, Davis, CA, USA

Rider control in bicycling is modeled by first adding the rider as a passive mechanism to the Whipple bicycle model. Next, for the rider control model a linear PID controller with or without delay is assumed, where the control inputs are the bicycle lean and steer angle with their higher derivatives, and the control output is the action-reaction steer torque applied by the rider at the handle bars. The experimental data is obtained from riding a bicycle on a narrow treadmill while applying an intermitted lateral perturbation by means of an impulse force applied at the seat post. The experiments are conducted in both the stable and the unstable forward speed range.  After some filtering, a parametric control model is fitted to the data. Finally, the gains of this control model are used to identify the specific LQR objective function which the rider is using to control the bicycle at the various forward speeds.

Stabilization control of a motorcycle during braking

S. Murakami1,, H. Nishimura1, and S. Zhu2

1 Keio University, Yokohama, Japan

2 Zhejiang University, Hangzhou, China


A rider maneuvers a motorcycle’s attitude stably by steering, shifting the weight of her/his body, accelerating or decelerating. Motorcycle overturning accidents often happen due to disturbances of uneven road or slippery road. In particular, the rider’s inappropriate operations largely affect the motorcycle motions and often cause the overturning accidents. Hence, a driving stability control system to assist the rider’s operations is necessary.

In this paper, a front-steering assist control is designed to stabilize a motorcycle during braking. The rider-motorcycle system with its pitching motions is linearized around an equilibrium point of quasi-steady state straight running with constant deceleration. From the viewpoints of eigenvalues and frequency responses, the linearized model is analyzed and a reduced-order model is obtained to design the control system by using H∞ control theory. By carrying out simulations during braking, it is demonstrated that the control system can stabilize a motorcycle when receiving a sudden disturbance from the front wheel and is robust against parameter variations and several braking situations.

Methods to assess the stability of a bicycle/rider system

Adrian Cooke1, Vera Bulsink1, Rosemary Dubbeldam2, Bart Koopman1, Wim Poelman1, Marc Beusenberg1 and Maarten Bonnema1

1Engineering Technology, University of Twente

2Roessingh Research and Development, Roessinghsbleekweg 33b,

The SOFIE (Intelligent Assisted Bicycles) project wishes to create performance and design guidelines for mechatronic appliances which improve the stability of electric bicycles, a so-called intelligent stability assist device (IAD). To achieve this goal, a stability hypothesis, an advanced rider/bicycle model and bicycle stability test bench, will be created. This paper describes the development of the stability hypothesis, the bicycle/rider model and the bicycle stability test bench. The stability hypothesis is based on the concept that the centre of mass of the bicycle/rider system stays within certain margins from the heading of a bicycle. The rider/bicycle model is created in the MSC software ADAMS for multi-body dynamic simulations. The bicycle stability test bench can be used for many different types of bicycles. The model and the test bench in combination with the stability hypothesis will be used to validate the effectiveness of the IAD's and assist in their design.

Why some Parkinson’s disease patients cannot stand or walk but can ride a bicycle - A control system-based analysis

Jingang Yi1, Damoon Soudbakhsh2, Yizhai Zhang1, and Yang Zhang1

1Dept. of Mechanical and Aerospace Engineering, Rutgers University

2Dept. of Mechanical Engineering, Massachusetts Institute of Technology

Recent clinic studies report a set of striking observations that some Parkinson’s disease (PD) patients, who cannot maintain a stance balance and walk (called freezing of gait (FOG)), are able to freely ride bicycles [1–3]. Bicycling is also reported as a possible clinical diagnosis tool for atypical Parkinsonism [4]. From the pathophysiology viewpoint, several hypotheses are  proposed to explain why the FOG and PD patients cannot balance and walk but yet being able to maintain the bicycling capability [2]. These hypothetical explanations range from multi-modal sensory cue differences in walking and bicycle, different defective cortical areas for walking and bicycling motor skills, dynamic balance differences between walking and bicycling, etc. This remarkably preserved ability to balance bicycles also implies possible therapeutic values of bicycle-assisted rehabilitation for FOG and PD patients [2, 5].

The goal of this paper is to present an explanation of the above-discussed preserved bicycling ability for FOG and PD patients from a control system viewpoint. The rationale of taking control systems viewpoint to analyze balance stability of the human rider/bicycle systems is twofolds. First, human sensorimotor skills, especially the balance control in the frontal plane, can be modeled as a time-delay feedback system [6,7]. Such a neurophysiological modeling approach anbles us to take advantage of well-developed feedback control theory to analyze the complex rider/bicycle interactions. Second, the analysis of the integrated human sensorimotor control with the bicycle dynamics provides a rigorous and powerful tool not only to possibly explain the preserved bicycling ability for FOG and PD patients but also to guide the design of the bicycle-assisted rehabilitation. The purpose of this paper is not to validate any pathophysiological hypotheses discussed in literature but rather to provide the possible explanation from the control-theoretic viewpoint. Therefore, the analysis and discussion in this paper complement the existing hypotheses. Comparing with our previous work in [8], we extend the stability analysis of the human motor control and present new parameter sensitivity study. We also add more experimental validation of human control torque model in this paper.

[1] Snijders, A. H., and Bloem, B. R., 2010. “Cycling for freezing of gait”. New Engl. J. Med., 362, p. e46.

[2] Snijders, A. H., Toni, I., Ru˙ziˇ cka, E., and Bloem, B. R., ˇ 2011. “Bicycling breaks the ice for freezers of gait”. Mov. Disord., 26(3), pp. 367–371.

[3] Stamelou, M., Kojovic, M., Edwards, M. J., and Bhatia, K. P., 2011. “Ability to cycle despite severe freezing of gait in atypical Parkinsonism in Fahr’s syndrome”. Mov. Disord., 26(11), pp. 2141–2142.

[4] Aerts, M. B., Abdo, W. F., and Bloem, B. R., 2011. “The “bicycle sign” for atypical parkinsonism”. Lancet, 377, pp. 125–126.

[5] Song, C.-G., Kim, J.-Y., and Kim, N.-G., 2004. “A new postural balance control system for rehabilitation training based on virtual cycling”. IEEE Trans. Inform. Technol. Biomed., 8(2), pp. 200–207.

[6] Goodworth, A. D., and Peterka, R. J., 2009. “Contribution of sensorimotor integration to spinal stabilization in humans”. J. Neurophysiol., 102, pp. 496–512.

[7] Goodworth, A. D., and Peterka, R. J., 2010. “Influence of bilateral vestibular loss on spinal stabilization in humans”. J. Neurophysiol., 103, pp. 1978–1987.

[8] Soudbakhsh, D., Zhang, Y., and Yi, J., 2012. “Stability analysis of human balance control of stationary bicycles”. In Proc. Amer. Control Conf.

Steering neuromuscular dynamics: motorcycle riders vs car drivers

M. Massaro1 and D.J. Cole2

1Department of Industrial Engineering, University of Padova

2Department of Engineering, University of Cambridge

Few papers on the neuromuscular interaction of the driver with the steering system while driving four-wheeled vehicles or two-wheeled vehicles have been published so far. Nevertheless the neuromuscular dynamics affect the response of the driver-vehicle system and in the case of single-track vehicles significant effect on stability are documented [1]. In addition, the theoretical understanding of human steering control behavior is essential when using numerical simulation to develop sophisticated systems that actively modify the steering angle and torque in four-wheeled vehicles [2].

One of the earliest attempts to understand the role of neuromuscular dynamics in the driver–vehicle–steering system dated back to 1993, [1]. In 2004 a fixed-base car driving simulator with a torque feedback steering wheel was used to identify the passive properties of the driver’s arms [4]. The experiments involved the test subject holding the steering wheel with both hands whilst a random torque demand signal was sent to the steering system. Subsequently, the stiffness and damper values of the driver-steering wheel system identified on the car simulator were used to model the driver-handlebar system in motorcycle [5], and some consideration on the vehicle stability were addressed. It should be noted that different muscles are involved when riding a motorcycle, however no motorcycle related data were published at the time. In 2007 [6] further results on the dynamic properties of a driver's arms holding a steering wheel were reported. In 2011 the identification of the driver-handlebar properties were identified on a motorcycle simulator [7]. The experiments involved the test subject holding the handlebar with both hands whilst a sine swept signal was sent to the steering system. The identified driver properties were used to evaluate the motorcycle stability and compared with road tests giving good correlation [1].

This paper addresses the comparison between the most recent findings on motorcycle rider and car driver properties. In addition, the role of neuromuscular dynamics in the stability of the motorcycle-rider and car-drivers systems is discussed.

[1] Massaro M., Lot R., Cossalter V., Brendelson J., Sadauckas J., 2011, Numerical and experimental investigation on the significance of rider on motorcycle weave, 22nd IAVSD, 14-19 August, Manchester Metropolitan University.

[2] Cole D. J., 2012, A path-following driver–vehicle model with neuromuscular dynamics, including measured and simulated responses to a step in steering angle overlay, Vehicle System Dynamics, in press.

[3] Modjtahedzadeh A. and Hess R.A., 1993, A model of driver steering control behaviour for use in assessing vehicle handling qualities, ASME J. Dyn. Syst. Meas. Control 115, pp. 456–464.

[4] Pick AJ and Cole D.J., 2004, Neuromuscular dynamics and the vehicle steering task, Vehicle System Dynamics, 41(Suppl):l82–l91

[5] Sharp RS, Limebeer DJN, 2004, On steering oscillations of motorcycles. Proc of IMechE, Part C, Journal of Mechanical Engineering Science, 218:1449–1456

[6] Pick A J and Cole D J, 2007, Dynamic properties of a driver's arms holding a steering wheel, Proc of IMechE, Part D: Journal of Automobile Engineering, 221: 1475

[7] Cossalter V., Doria A., Lot R., Massaro M., 2011, The effect of rider’s passive steering impedance on motorcycle stability: identification and analysis, Meccanica, 46: 279–292

Closed form and numerical solutions of traction in cycling

Junghsen Lieh, PhD

Professor, Mechanical & Materials Engineering

Wright State University, Dayton, Ohio 45435

In theory, how fast can a cyclist pedal and how much power is required from him? Traditionally, pure numerical methods have been used to answer these questions. To improve this, a closed-form method is developed and used to solve for the speed and power requirement during cycling and its results are compared with those obtained from numerical methods. The simulation is conducted in Matlab/Simulink and used to study the effect of air drag coefficient, frontal area, rolling resistance and road gradient on the speed and power consumption.