When a bicycle rider moves into a bumpy terrain, he/she would change to a standing posture to take advantage of the visco-elastic knee dynamics to stabilize both the body and the bicycle. Similarly, while climbing up stairs holding a mug full of coffee, one would bend the elbow to let its visco-elastic dynamics stabilize the mug. In both cases, the internal impedance control strategy would have benefited from some generalization of previous experience in similar environments.
This project investigates how a robot interacting with an uncertain environment could passively migrate from one state to another just by changing its internal impedance. The paper describing how a simple walker known as the rimless wheel could use internal memory primitives of steady state dynamics recorded at different internal visco-elastic parameter values to passively move in the state space was presented at IROS2012.
Another paper presented how a robotic gripper holding an uncertain visco-elastic object could use a real time estimate of the probability of losing the grip to maintain grip on the uncertain object.
A paper presented in ICRA2012 showed the dominant sources of variability of passive dynamic walking. We observe steady state variability even on a seemingly deterministic hard ramp. We showed using experimental evidence and numerical simulations that, the interaction effect between the distribution of the coefficient of restitution and the coefficient of friction plays a dominant role in determining the steady state variability of walking. The initial velocity influences steady state variability at certain high velocities and high average coefficient of friction. Therefore, out of the dominant sources, the interaction between the velocity of walking and the coefficient of friction demonstrates an internal integration effect on the steady state variability. The movie clip of experiments is shown below:
Since the coefficient of restitution depends on the impedance of both the ground and the leg, internal impedance control of the legs will be a good strategy to keep the steady state variability of walking within desired bounds. We propose soft robots with controllable stiffness as a potential solution that can adjust the internal impedance to maintain desired steady state behaviors.
Soft continuum appendages exhibit high inherent compliance and stiffness modulation than their skeletal counter parts. Thus, they can tolerate wider range of impacts without compromising either safety or stability. For instance, these soft arms are safer to operate in unpredictable human environments. Due to their deformable arms/bodies they can overcome/penetrate further/deeper than the rigid bodied robots i.e., during search and rescue mission amidst building rubble.