Course Diagram
The Challenge
The “W” Prize of at least US$200,000 is offered for successful demonstration of unprecedented
efficiency and dexterity in machine locomotion.  It is intended to stimulate advances in the state of the
art, and to focus energy upon practical applications.  Additional objectives are to attract public attention,
and to encourage participants to have some good fun.  The rules are designed to exercise walking
machines, but if another type of machine is better able to complete the task, then so be it!

The essence of the task is to travel 10 km in no more than 10,000 seconds while using no more than 10
kJ per kg of machine mass - and occasionally surmounting obstacles which would stand in the way of
ordinary wheeled vehicles, but could easily be negotiated by a pedestrian.  The obstacles are shown in
Figure 1.  The course is designed to be economically replicable at most universities worldwide, so that
entrants may compete for the Prize at the time and place of their choosing.  Obstacles are therefore
made more sparse and regular than on truly rough terrain, and some rules are imposed to prevent
competitors taking undue advantage of artificiality.

Interpretation of the rules, specification of additional rules, and adjudication of attempts on the Prize
will be at the sole discretion of a
committee of three referees.  The referees will hold the Prize money in
an escrow account, accumulating interest and contributions from sponsors, until the Prize is awarded.

The course has four sets of obstacles arrayed along a 100 metre track.  These are as follows:

1. Size-limiting arch
The course begins and ends with an arch.  Its purpose is to exclude large machines such as “monster
trucks”, and instead limit size to something comparable to a person.  An entrant machine must prepare
for an attempt by entering the arch without external assistance, and position itself (in any orientation) at
rest while contained entirely within the arch.  At the end of each out-and-back circuit of the course, the
machine must completely exit the arch, reverse direction, and re-enter.  As a demonstration of dexterity,
the machine must make each reversal in no more than 4 seconds, as timed from breaking the exit plane
of the arch to breaking the same plane on re-entry.

2. Elevated stepping stones
The machine must negotiate a series of twenty stepping stones.  These are intended to test delicacy as
well as dexterity.  Hence they must rest on the track without external supports or fastenings.  A machine
may move or knock down blocks, so long as it uses only upright blocks for support while between the
1st and 20th blocks in the set (which must remain standing throughout).  Each pass over the stepping
stones must be completed in not more than 25 sec, measured from making contact with the 1st block to
breaking contact with the 20th block.

Now-you-see-it-now-you-don’t ditch
As a test of robustness to disturbances, the machine must cross a set of three slightly elevated panels,
the centre one of which will be removed (while the machine is elsewhere) on a randomly-selected set of
half of the passes through the course.  An entrant will not be informed of the remove-and-replace
sequence chosen for any given attempt on the Prize.

On at least one pass in five (i.e. once per kilometre traveled) the machine must climb a flight of stairs,
pass completely through the stair-exit plane shown in
the course diagram while on or above the upper
landing, and descend the stairs back onto the track.  This task is required on only a few circuits in order
to prevent stair climbing from contributing excessively to energy consumption.  On other circuits, the
machine may avoid the stairs, so long as it passes completely through the stair-exit plane.  In either case,
the machine is allowed a maximum of 10 seconds between crossings of the plane through the lower step.

The arch and staircase are obviously located at opposite ends of the course.  Location of the stepping
stones and the ditch is not critical; centering each anywhere within 10 m of the ¼ and ¾ points along
the course is fine.  The location of the obstacles can be measured precisely prior to an attempt.  This
differential-GPS to be used for navigation if desired, rather than some sort of obstacle-detection
system (as used in the
DARPA Grand Challenge).  Also the course may be marked or fitted with
sensors as desired, provided that they do not provide energy to the machine.

Further rules
•  Energy content rule
The energy used is calculated for a battery as the integral of its voltage-discharge curve, and for
combustible fuels as the lower heating value of fuel consumed (e.g. 42.5 MJ/kg for gasoline).  Other
energy stores will be assessed by the referees.  The allowable energy budget will be determined by the
lower of machine masses measured before and after a successful attempt.

No-Batmobiles rule
The machine must use the same set of supports on all parts of the course.  (Thus for example it may
not roll on the flat bits, and then deploy arms or legs for the obstacles.)

•  Keep-your-hands-to-yourself rule
No person, nor any object other than the track or obstacles, may touch the machine during an attempt
on the Prize.  (However teleoperation is perfectly acceptable.)

•  Tom McMahon memorial rule
The stiffness of the track must be high,
i.e. the natural frequency of a brick of the same mass as a
contestant machine, oscillating in “heave” on the track, must be high compared with the machine’s gait
frequency (if applicable).

•  No-Frankenstein rule
No living tissue may be used as a functional component of the machine.

•   If-God-wanted-us-to-fly rule
A supporting surface of the machine must contact the ground at least once in any three-second interval.  
(Hence walking, running, jumping, and crawling are admissible, but flying is not.)

Participation in the competition
Prospective participants must submit a short report and video to the referees to indicate their readiness
for an attempt on the Prize.   The referees will then arrange for observation of the attempt, which may
be done by a designated local observer rather than by a referee.

While the course is designed to allow participation “at home”, we want also to encourage exciting head-
to-head competition at an agreed place and time.  One-upmanship is therefore allowed.  Thus if a
machine completes the course within the time and energy budget, it will be awarded a score of 1 point
for each second and for each J/kg under budget.  The first competitor to maintain a leading score for 72
hours will be awarded the Prize.

Autonomous-vehicle components
Entrants will require some onboard computing, and most likely a control station and wireless
communications link.  A lot of work would be required to develop everything from scratch, but this is
not necessary.  Entrants might for example adapt one of the systems that has been developed for
autonomous aircraft (
e.g. from Athena, Cloud Cap, Micropilot, or Geneva Aerospace).  Much of their
hardware and software could be used as-is for communications, navigation, and systems monitoring.  
Aircraft-specific systems however have the disadvantage of falling under the
International Traffic in
Arms Regulations, which limits availability outside the country of origin.  A better option might be to
find an ITAR-free system on which participants can choose to standardise, and share information in the
interest of mutual support.

The following plots shows energy costs for a variety of transport modes (cf. Gabrielli and von Karman
1950).  The specific cost of transport for each is defined as follows, with an accompanying calculation
for the Prize task.
The task calls for an improvement in efficiency of about twofold relative to that achieved by the Cornell
biped (
Collins et al 2005) which seems to be the most efficient legged machine built to date.  Substantial
room for such improvement remains.  The specific resistance for walking on level ground is calculated
to be only 0.02 or less (
McGeer 1992, 1993; Kuo et al 2005), so the target of 0.10 set for the Prize
leaves a substantial margin for energy-conversion inefficiencies, and for computing and communication
overheads.  Still, success in the W competition will set a new standard for land mobility (
cf. Tucker
1975) and, we hope, stimulate the advance of practical applications.
1. Collins, S.H., Ruina, A.L., Tedrake, R., and Wisse, M. (2005) Efficient bipedal robots based on
passive-dynamic walkers. Science 307: 1082-1085.

2. Gabrielli G., and von Karman T.  
What price speed?  Mech Eng. 1950;72:775–781

3. Kuo, A.D., Donelan, J.M., and Ruina, A. (2005)
Energetic consequences of walking like an inverted
pendulum: Step-to-step transitions. Exercise and Sport Sciences Reviews 33: 88-97.

4. McGeer, T.  
Principles of walking and running. In Advances in Comparative and Environmental
Physiology 11: Mechanics of Animal Locomotion. R. McN. Alexander, ed. Berlin: Springer-Verlag 1992.

5. McGeer, T.  
Dynamics and control of bipedal locomotion. Journal of Theoretical Biology 163, 277-
314, 1993.

6. Tucker, V.A.  The Energetic Cost of Moving About.  
American Scientist, July-August 1975.
Course Diagram