Evaluation of low-power devices for smart greenhouse development
The combination of artificial intelligence and the Internet of Things (AIoT) is enabling the next economic revolution in which data and immediacy are at the key players. Agriculture is one of the sectors that can benefit most from the use of AIoT to optimise resources and reduce its environmental fo...
| Autores: | , , , , , |
|---|---|
| Formato: | artículo |
| Fecha de publicación: | 2023 |
| País: | España |
| Recursos: | Universidad Católica San Antonio de Murcia (UCAM) |
| Repositorio: | RIUCAM. Repositorio Institucional de la Universidad Católica San Antonio de Murcia |
| OAI Identifier: | oai:repositorio.ucam.edu:10952/7385 |
| Acesso em linha: | http://hdl.handle.net/10952/7385 |
| Access Level: | acceso abierto |
| Palavra-chave: | Artificial Intelligence Edge computing Time-series forecast TinyML CPU-GPU Performance Power consumption |
| Resumo: | The combination of artificial intelligence and the Internet of Things (AIoT) is enabling the next economic revolution in which data and immediacy are at the key players. Agriculture is one of the sectors that can benefit most from the use of AIoT to optimise resources and reduce its environmental footprint. However, this convergence requires computational resources that enable the execution of AI workloads, and in the context of agriculture, ensuring autonomous operation and low energy consumption. In this work, we evaluate TinyML and edge computing platforms to predict the indoor temperature of an operational greenhouse in situ. In particular, the computational/energy trade-off of these platforms is assessed to analyse whether their use in this context is feasible. Two artificial neural networks (ANNs) are adapted to these platforms to predict the indoor temperature of the greenhouse. Our results show that the microcontroller-based devices can offer a competitive and energy-efficient computational alternative to more traditional edge computing approaches for lightweight ML workloads. |
|---|