Neural network design for energy-autonomous artificial intelligence applications using temporal encoding

Abstract

<jats:p>Neural networks (NNs) are steering a new generation of artificial intelligence (AI) applications at the micro-edge. Examples include wireless sensors, wearables and cybernetic systems that collect data and process them to support real-world decisions and controls. For energy autonomy, these applications are typically powered by energy harvesters. As harvesters and other power sources which provide energy autonomy inevitably have power variations, the circuits need to robustly operate over a dynamic power envelope. In other words, the NN hardware needs to be able to function correctly under unpredictable and variable supply voltages. In this paper, we propose a novel NN design approach using the principle of pulse width modulation (PWM). PWM signals represent information with their duty cycle values which may be made independent of the voltages and frequencies of the carrier signals. We design a PWM-based perceptron which can serve as the fundamental building block for NNs, by using an entirely new method of realizing arithmetic in the PWM domain. We analyse the proposed approach building from a 3 × 3 perceptron circuit to a complex multi-layer NN. Using handwritten character recognition as an exemplar of AI applications, we demonstrate the power elasticity, resilience and efficiency of the proposed NN design in the presence of functional and parametric variations including large voltage variations in the power supply.</jats:p> <jats:p>This article is part of the theme issue ‘Harmonizing energy-autonomous computing and intelligence’.</jats:p>

Modeling the motion processes of a multifunctional robot for animal units

“…In case when the design parameters of the model are known, the forces can be determined by using formulas (13-15), as well as the moments acting on each wheel. Thus, the right-hand sides of the system of equations (5) are calculated. The set of equations (7-9, 13, 14) is a dynamic model of a robot with a drive in the form of two independent wheels.…”

“…These forces must be taken into account when drilling the rotation of the developed wheeled robot on rubber mats mounted on the surface of the serviced space inside the animal unit. According to the studies of Mileiko, S., Bunnam, T., Xia, F., Shafik, R., Yakovlev, A., & Das, S. [5], it was found that the creation of artificial neural systems for applications of artificial intelligence requires considering physical parameters of a controlled object [5], in the role of which a wheeled robot acts as part of our research.…”

Simulation of the dynamic processes of a low-capacity combine harvester movement

“…There are a lot of orientation methods, as well as components for these purposes, and to choose the optimal control method, one must proceed from the existing conditions. One of the field-acceptable methods of orientation in space is terrain mapping, which is an important component of the introduction of accurate farming methods [8]. Promising is the use of an autonomous control system based on laser scanners that can navigate the terrain in real time [9].…”