What The Bicycle Can Tell Us About Human Control ¶

Jason K. Moore

Assistant Professor, TU Delft

UC Davis First Year Seminar: Bicycle Engineering

November 4, 2020

In 2004, I wanted to be¶

Sam Whittingham, world record (~80 mph)

Varna passing by¶


Your browser does not support the video tag.

Evie Before: 6:49 AM¶

Evie After: 6:50 AM¶

As a start, we engineered this¶

But it was quite hard to balance¶


Your browser does not support the video tag.

With practice you can learn¶


Your browser does not support the video tag.

Questions raised¶

  • Why is one bicycle harder to control than another?
  • How can we design a bicycle such that it is easy to control?

Little did I know I'd be thinking about this for 15+ years¶

A little about bicycle dynamics¶

Bicycles are curious machines

  • Bicycles are the most efficient from of human transportation (lowest energy per person per distance)
  • In general, you have to balance a bicycle before directing it from point A to B
  • It is possible to balance a bicycle without touching the handlebars
  • People can do all kinds of stunts on bicycles

Bicycle Balet¶


Extreme Sports¶


Your browser does not support the video tag.

Question¶

If you want you bicycle to turn to the left, what direction do you turn the handlebars?

Countersteering video¶


Your browser does not support the video tag.

Why do you have to countersteer?¶

If you are in a steady right hand turn, you must be leaning to the right. So, the only way to initiate a right hand lean is to move the wheel contact points out to the left.

Think about balancing a broom.

Newton's Laws are able to predict the need to countersteer¶


Your browser does not support the video tag.

Another curiosity¶


Your browser does not support the video tag.

Bicycles can be stable¶

Also predicted by Newton's Laws!

$F = m a$

Most bicycles are stable once they reach a certain speed, i.e. if you try to knock it over it rights itself.

Bicycle at $v=0$¶


Your browser does not support the video tag.

Bicycle at $v=3$ m/s¶


Your browser does not support the video tag.

Bicycle at $v=4$ m/s¶


Your browser does not support the video tag.

Bicycle at $v=6$ m/s¶


Your browser does not support the video tag.

What makes a bicycle stable?¶

Gyroscopic Action¶

"The wheel is a gyroscope so it always steers in the correct direction"

Caster Trail¶

"The front wheel of a bicycle trails behind the steering axis just like a caster on an office chair, so the wheel always corrects the steering."

What if gyroscopic action is removed?¶

What if gryoscopic action and caster are removed!¶

TU Delft Two Mass Skate Bicycle

TMS Video¶


What if gyroscopic action is increased?¶


Your browser does not support the video tag.

What happens when human is in control?¶


Your browser does not support the video tag.

Human Sensory Feedback¶

Sensors¶

  • We sense our body position, orientation, and configuration with our eyes
  • We sense our body's configuration with muscle spindles
  • We sense our orientation wrt to gravity, orientation rate, and angular acceleration with the utricle in our ear
  • We sense forces being applied to our body with our proceptive system

Limitations¶

  • Time delays from sensing to acutation
  • Limits in acutation forces
  • Limits in frequency of motion

Use of instrumented bicycles¶

Experiments¶

Tracking Experiments¶


Your browser does not support the video tag.

Perturbation Response Experiments¶


Your browser does not support the video tag.

Mathematical models of human control¶

Feedforward and feedback mathematical models can predict the human sensing and actuation relationship.

Why does any of this matter?¶

  1. The dynamics of the bicycle describe how it moves when forces are applied to it.
  2. Humans apply forces to a bicycle to make it transport them where they want to go.
  3. The dynamics can be changed by changing the vehicles physical properties, i.e. mass, geometry, tires, etc.

So, there must be a link between the physical properties of the vehicle and how easy it is for the human to control it.¶

This leads to my research interests:

How do we design human controlled machines (e.g. vehicles) such that they perfectly compliment the human's intentions?

The approach¶

  • Create mathematical models that predict how machines move if forces are applied to them.
  • Create mathematical models that predict how a person will actuate their muscles to complete the tasks if they sense certain cues.
  • Design optimal machines with optimal control systems such that the human can easily, efficiently, and accurately control the machine.

Optimal Bicycle Designs¶

Design can be treated as an optimization.

Once you have mathematical definitions of a system you can then use mathematical and computational techniques to answer questions like:

"What would the geometry of the vehicle be if it was the easiest to control?"

Example Optimal Geometry¶

Geometries for Different Speeds¶

Other Applications¶

Steer Assist Bicycle¶

Steer-By-Wire Bicycle¶

Self-balancing Electric Motorcycles¶


BMW's Concept

More info¶

  • My website: http://www.moorepants.info
  • Our lab website: http://mechmotum.github.io
  • My 2012 dissertation: http://moorepants.github.io/dissertation/
  • Papers: http://tinyurl.com/jkm-gscholar
  • Twitter, G+, Linkedin, etc: moorepants
  • My email: j.k.moore@tudelft.nl
  • Link to slides: https://www.moorepants.info/presentations/2020/ucd-fys-bicycle