-
Quantized deep learning models on low-power edge devices for robotic systems
Abstract: In this work, we present a quantized deep neural network deployed on a low-power edge device, inferring learned motor-movements of a suspended robot in a defined space. This serves as the fundamental building block for the original setup, a robotic system for farms or greenhouses aimed at a wide range of agricultural tasks. Deep learning on edge devices and its implications could have a substantia… ▽ More
Submitted 30 November, 2019; originally announced December 2019.
Comments: Presented at NeurIPS 2019 Workshop on Machine Learning for the Develo** World