|
 |
Nonlinear Control, Legged Robot
Numerous attempts have been made by scholars to realize a biped walking
robot.
Most of them focused on a walking motion, and they tried to come
true the walking motion on various environments.
By the way, the motivation of the study for the biped walking robot
include that ``to realize a useful humanoid robot'' in some sense.
Thus we request the robot to be able to carry out the task efficiently
by similar human motions that someone should do.
In the sense, it is important that we realize the biped walking motion
as a human movement faculty.
We shall now look into a conveyance which looks easy.
In this case, it is clear that the robot must accomplish the following
four phases at least; [Phase 1] The robot finds the load, and approaches the target (stationary walking).
[Phase 2] It sits down to hold the target (non-stationary motion).
[phase 3] It stands up holding the load (non-stationary motion $+$ dynamics variation).
[Phase 4)] It walks to the destination with the load (stationary walking $+$ dynamics variation).
However it can be said that the almost traditional researches on biped
robots mentioned above are only concerned with the stationary walking.
We notice that the traditional biped robot cannot carry out the task that
looks easy.
Hence, it is important that non-stationary motions which complement the
traditional studies on the biped walking robot are realized.
Moreover we need to analyze stability of the motion.
Of course, for the biped robot whose foot bottom is enough broad, we can
propose a control method that each joint trace a scheduled trajectory to
realize non-stationary motions theoretically.
In this case, we guarantee stability in the sense that the robot dose
not fall down with a sufficient ankle torque.
Generally speaking, the ankle torque has a constraint, as the bottom of
foot is relatively small to the body.
Therefore, the control input may violate the constraints if we design the
control scheme disregarded the constraints.
Thus, it is important to discuss the ankle torque constraint.
From the motivation, we considered realizing the non-stationary motion
through two points of view.
In the first place, we thought over the case that the non-stationary
motion did not use the ankle torque so that we could avoid the ankle
torque constraint.
As the result, we designed a non-linear gain scheduling control
scheme to realize the non-stationary motion.
Furthermore, we verified the effectiveness of the control scheme through
some simulations and experiments.
In the second place, we considered the case that the non-stationary
motion used the ankle torque under the constraint.
Especially, we focused on a standing upright motion as a special case of
the non-stationary motion.
As the result, we proposed a hybrid control design method so that we
could take the ankle torque constraint into consideration explicitly.
Moreover we showed the effectiveness of the control scheme through a
simulation.
Research papers (publications of university or research institution)
