Control in technological systems and physical intelligence: an emerging theory

The transduction and processing of physical information is becoming important in a range of research fields, from the design of materials and virtual environments to the dynamics of cellular microenvironments. Previous approaches such as morphological computation/soft robotics, neuromechanics, and embodiment have provided valuable insight. This work approaches haptic, proprioception, and physical sensing as all part of the same subject. In this presentation, three design criteria for applying physical intelligence to engineering applications will be presented. These criteria have several properties in common, which inspires two types of end-effector model: stochastic (based on a spring) and deterministic (based on a piezomechanical array). The generalized behavior and output dynamics of these models can be described as three findings summarized from previous work. In conclusion, future directions for modeling neural control using a neuromorphic approach will be discussed