By John Blankenship
and Samuel Mishal
The long-range goal for many robot hobbyists is the development of a
humanoid robot, especially a walking humanoid. The good news is there
are many humanoid robots available — both as kits and assembled units.
The bad news is that most of the walking robots available — especially
those with reasonable prices — are not truly suitable for learning about
robotic walking algorithms.
Programming with Poses
The software that comes with many humanoid robots
lets the user “program” the robot by creating a variety
of poses and then linking those poses together in an
appropriate sequence to create lifelike motions. This
methodology has the advantage of allowing nearly anyone
to quickly create custom demonstrations such as walking or
dancing that make the robot appear to have some degree
of intelligence. Unfortunately, robots programmed in this
way are far less capable than they seem.
A simulation of a
walking humanoid robot.
38 SERVO 11.2009
Robots that walk using a sequence of poses, for
example, often cannot handle even simple problems such
as gentle inclines or uneven terrain, let alone steep
grades or stairways. Robots capable of dealing with those
situations require more sensors than are available on the
humanoid robots currently aimed at the hobbyist market.
In many cases, the only sensory information available from
a walking robot is the angle of each of the joints.
A proper walking robot must know the location of its
limbs, when it is falling, the direction of the fall, and when
each foot makes contact with the ground. Based on our
experimentation, this information is the bare minimum
needed to create a walking algorithm.
In order to make it easy to experiment with walking
algorithms, a simple walking robot simulator was developed
using RobotBASIC because of its ability to create flicker-free
animation. Figure 1 shows the humanoid figure created
by the simulation. A major goal for the simulator was to
maintain simplicity so that programmers new to walking
algorithms could be productive with a reasonable amount
of effort. Because of that, some compromises had to