Mechanical analysis of avian feet: multiarticular muscles in grasping and perching.

The grasping capability of birds' feet is a hallmark of their evolution, but the mechanics of avian foot function are not well understood. Two evolutionary trends that contribute to the mechanical complexity of the avian foot are the variation in the relative lengths of the phalanges and the subdivision and variation of the digital flexor musculature observed among taxa. We modelled the grasping behaviour of a simplified bird foot in response to the downward and upward forces imparted by carrying and perching tasks, respectively. Specifically, we compared the performance of various foot geometries performing these tasks when actuated by distally inserted flexors only, versus by both distally inserted and proximally inserted flexors. Our analysis demonstrates that most species possess relative phalanx lengths that are conducive to grasps actuated only by a single distally inserted tendon per digit. Furthermore, proximally inserted flexors are often required during perching, but the distally inserted flexors are sufficient when grasping and carrying objects. These results are reflected in differences in the relative development of proximally and distally inserted digital flexor musculature among 'perching' and 'grasping' taxa. Thus, our results shed light on the relative roles of variation in phalanx length and digit flexor muscle distribution in an integrative, mechanical context.

A cellular control architecture for compliant artificial muscles.

Dividing an artificial muscle material into a network of small cells could provide performance benefits and eliminate unwanted behaviors such as hysteresis. This paper presents a scheme for the position control or compliance control of an artificial muscle having this kind of cellular structure. Each cell contracts or relaxes probabilistically in response to a global feedback control loop, which measures only the aggregate force and displacement of the muscle. The stochastic nature of the cells produces smooth, reliable global behavior in the artificial muscle. By choosing a control law such that the expected response of the artificial muscle is equal to the desired response, good tracking control is achieved.