A Neuromorphic Architecture for Scalable Event-Based Control

Discrete methodologies use the discrete language of finite state machines while continuous methodologies use the continuous modelling language of differential equations. The resulting need to interface discrete automata and continuous differential equations has made the theory of hybrid and cyberphysical systems a key component of control design. The advantages of separating the discrete and the continuous design are many. Yet, this separation has become a bottleneck in the design of scalable architectures that can control and regulate across a range of spatial and temporal scales [1, 2]. The divide between automation and regulation does not exist in animal machines. Nervous systems make decisions and regulate muscle activation with one and the same architecture that is event-based rather than continuous or discrete.