The field of neuromorphic control and excitable systems is witnessing significant advancements, with a focus on developing innovative methods for robustness verification, stability analysis, and control of complex systems. Researchers are exploring the potential of neuromorphic architectures, such as spiking neural networks and excitable circuits, to create low-latency, low-energy, and adaptive control systems. Notable developments include the use of novel approximation techniques for nonlinear activation functions, emulation-based design procedures for stabilizing linear time-invariant systems, and the introduction of new stability notions, such as integral spiking-input-to-state stability. Furthermore, researchers are investigating the application of discrete-event modeling and scenario-aware control to optimize the performance of neuromorphic circuits and systems. Some noteworthy papers in this area include:
- A paper proposing a novel truncated rectangular prism approximation for RNN robustness verification, which demonstrates significant improvement over state-of-the-art approaches.
- A paper presenting a new systematic approach for stabilizing linear time-invariant systems using spiking neural network-based controllers, which establishes a certifiable practical stability property of the closed-loop system.