Neuromorphic Robust Estimation of Nonlinear Dynamical Systems Applied to Satellite Rendezvous
Authors:
Reza Ahmadvand,
Sarah Safura Sharif,
Yaser Mike Banad
Abstract:
State estimation of nonlinear dynamical systems has long aimed to balance accuracy, computational efficiency, robustness, and reliability. The rapid evolution of various industries has amplified the demand for estimation frameworks that satisfy all these factors. This study introduces a neuromorphic approach for robust filtering of nonlinear dynamical systems: SNN-EMSIF (spiking neural network-ext…
▽ More
State estimation of nonlinear dynamical systems has long aimed to balance accuracy, computational efficiency, robustness, and reliability. The rapid evolution of various industries has amplified the demand for estimation frameworks that satisfy all these factors. This study introduces a neuromorphic approach for robust filtering of nonlinear dynamical systems: SNN-EMSIF (spiking neural network-extended modified sliding innovation filter). SNN-EMSIF combines the computational efficiency and scalability of SNNs with the robustness of EMSIF, an estimation framework designed for nonlinear systems with zero-mean Gaussian noise. Notably, the weight matrices are designed according to the system model, eliminating the need for a learning process. The framework's efficacy is evaluated through comprehensive Monte Carlo simulations, comparing SNN-EMSIF with EKF and EMSIF. Additionally, it is compared with SNN-EKF in the presence of modeling uncertainties and neuron loss, using RMSEs as a metric. The results demonstrate the superior accuracy and robustness of SNN-EMSIF. Further analysis of runtimes and spiking patterns reveals an impressive reduction of 85% in emitted spikes compared to possible spikes, highlighting the computational efficiency of SNN-EMSIF. This framework offers a promising solution for robust estimation in nonlinear dynamical systems, opening new avenues for efficient and reliable estimation in various industries that can benefit from neuromorphic computing.
△ Less
Submitted 14 May, 2024;
originally announced May 2024.
A Cloud-Edge Framework for Energy-Efficient Event-Driven Control: An Integration of Online Supervised Learning, Spiking Neural Networks and Local Plasticity Rules
Authors:
Reza Ahmadvand,
Sarah Safura Sharif,
Yaser Mike Banad
Abstract:
This paper presents a novel cloud-edge framework for addressing computational and energy constraints in complex control systems. Our approach centers around a learning-based controller using Spiking Neural Networks (SNN) on physical plants. By integrating a biologically plausible learning method with local plasticity rules, we harness the efficiency, scalability, and low latency of SNNs. This desi…
▽ More
This paper presents a novel cloud-edge framework for addressing computational and energy constraints in complex control systems. Our approach centers around a learning-based controller using Spiking Neural Networks (SNN) on physical plants. By integrating a biologically plausible learning method with local plasticity rules, we harness the efficiency, scalability, and low latency of SNNs. This design replicates control signals from a cloud-based controller directly on the plant, reducing the need for constant plant-cloud communication. The plant updates weights only when errors surpass predefined thresholds, ensuring efficiency and robustness in various conditions. Applied to linear workbench systems and satellite rendezvous scenarios, including obstacle avoidance, our architecture dramatically lowers normalized tracking error by 96% with increased network size. The event-driven nature of SNNs minimizes energy consumption, utilizing only about 111 nJ (0.3% of conventional computing requirements). The results demonstrate the system's adjustment to changing work environments and its efficient use of computational and energy resources, with a moderate increase in energy consumption of 27.2% and 37% for static and dynamic obstacles, respectively, compared to non-obstacle scenarios.
△ Less
Submitted 12 April, 2024;
originally announced May 2024.
