Human Stiffness Behaviour
In the earliest studies of human arm impedance it was shown that the fundamental spring-like properties of the neuro-musculo-skeletal system generalise to multiple dimensions such that displacing the hand with a robot generates restoring forces toward the initial position. A fundamental question therefore is that of whether humans modulate limb stiffness in order to optimise performance depending on task constraints. Work toward understanding this question has been carried out on three fronts \x96 methods, behaviour and neural mechanisms \x96 during the second year of the STIFF project:
Methods: TUD has implemented methods to extract time-varying impedance values based on a new \x93linear parameter varying\x94 (LPV) algorithm originally developed for cyclic behaviour of wind turbines, while DLR has evaluated the use of their light weight robot arms as a means to perturb a human arm and measure stiffness in three dimensions. DLR has also developed a hand-held device to measure the stiffness of the human grip during manipulation tasks and are developing methods to map EMG to stiffness in an effort to measure human stiffness behaviour without having to perturb the arm.
Behavior: UPD and TUD have tested experimentally whether the principle of \x91minimum intervention\x92 governs the regulation of limb impedance for targeted movements of the hand. Whereas reducing the sensitivity of the limb to noise by having the hand move along a rigid constraint induced impedance changes consistent with this principle, increasing the tolerance for noise by relaxing the accuracy demands at the target did not. These human data are being modelled by UEDIN and IDSIA in the context of optimal feedback control to try and understand the performance criteria or \x91cost-functions\x92 that the CNS uses to optimize impedance in these situations.
Mechanisms: UPD is using physiological techniques to study the neural basis for human stiffness behaviour. They have looked at the sensitivity of the motoneuron to proprioceptive feedback from the muscle using the H-reflex technique, to clarify the relationship between stretch reflex gain modulation and the level of muscle force or net applied force on the environment. They have also prepared experiments to test for a direct influence of cortical signals on the reversal of the stretch reflex during catching, by means of TMS, and to test whether muscles act as natural wave-variable processors to counteract significant transmission delays in the neural feedback circuits.