Meeting Abstract

S7-2.1.1  Jan. 6  GENERAL PROPERTIES OF LOCOMOTOR CONTROL SYSTEMS PROCHAZKA, A.*; YAKOVENKO, S.; University of Alberta, Edmonton; University of Montreal, Montreal arthur.prochazka@ualberta.ca

Complex tasks such as locomotion involve the neural integration of sensory input, spinal pattern generation and predictive control. The contributions of these mechanisms have recently been investigated with neuro-biomechanical models (Yakovenko et al. Biol Cybern 90: 146-155, 2004; Ekeberg and Pearson J Neurophysiol 94: 4256-4268, 2005). These studies produced the following conclusions regarding locomotor control. The control of phase durations by the locomotor CPG may rely on relatively simple biasing controls involving persistent inward currents. The centrally generated pattern can operate through the intrinsic force-generating properties of limb muscles and mechanical coupling between the limbs, to generate stable locomotion even in the face of small variations in terrain. Stretch reflexes provide some additional force adjustment and a limited means of changing speed and posture. Larger speed and terrain adjustments require higher-level control strategies similar to fuzzy logic control. Global rules based on sensory input are required for movement selection, predictions about upcoming movements and overall balance. We will discuss the following propositions regarding locomotor control: 1. Sensory input is generally multivariate, complex and “noisy”. 2. Motor actuators are nonlinear and difficult to model precisely. However, these nonlinearities seem to allow control strategies in biological systems that would be inappropriate or unstable in linear systems, e.g. positive force feedback. 3. There are numerous ways of performing a sensorimotor task successfully. 4. Combinations of control strategies (PID, finite state, fuzzy logic, global targets) are more likely to control complex systems successfully than single strategies.