This video clip shows two SILO4 robots walking indoors. They perform a discontinuous free gait on irregular terrain.
This video clip shows
the active compliance control with force feedback programmed in the SILO4 legged robot.
The robot acts as a system with springs and dampers in every joint and feet weighing 7.5
kg each. Of course, feet weigh much less than that, and the springs and dampers do not
exist physically. The behavior is implemented by software in the control law, that is, the
robot's twelve dc motors are told to make the joints behave as if they had springs and
dampers and the feet weigh 7.5 kg each.
This is programmed by defining virtual masses, springs and damping coefficients and then
writing the subsequent equation that relates forces, position and velocities in the
motors' control law.
The active compliance control scheme permits the robot adapt more softly to the ground and
stick more firmly on the ground. It also makes the body move more smoothly by reducing leg
crushes that result from the inherent inaccuracy of position control. This "crush
problem" is typical of systems with close kinematic chains.
Click here to download the video clip (6 MB)
Silo4 following a chaging-crab trajectory on flat terrain with a continuous free gait
Previous experiment (see previous video) is repeated using the gait adaptation approach using the gait patterns for an external force of 20N, which reduces CG height and leg stroke to maximize the stability margin. In this experiment the robot tumbled down at a force of 66.3 N therefore the resulting gait increases its robustness to external disturbances to a 27 percent for this robot.
SILO4 walking uphill using a conventional active-compliance controller
SILO4 walking uphill using active compliance with stability compensation.
