Transfer and Online Reinforcement Learning in STT-MRAM Based Embedded Systems for Autonomous Drones
Other Computer Science
2019-05-16 v1 Hardware Architecture
Abstract
In this paper we present an algorithm-hardware codesign for camera-based autonomous flight in small drones. We show that the large write-latency and write-energy for nonvolatile memory (NVM) based embedded systems makes them unsuitable for real-time reinforcement learning (RL). We address this by performing transfer learning (TL) on metaenvironments and RL on the last few layers of a deep convolutional network. While the NVM stores the meta-model from TL, an on-die SRAM stores the weights of the last few layers. Thus all the real-time updates via RL are carried out on the SRAM arrays. This provides us with a practical platform with comparable performance as end-to-end RL and 83.4% lower energy per image frame
Cite
@article{arxiv.1905.06314,
title = {Transfer and Online Reinforcement Learning in STT-MRAM Based Embedded Systems for Autonomous Drones},
author = {Insik Yoon and Aqeel Anwar and Titash Rakshit and Arijit Raychowdhury},
journal= {arXiv preprint arXiv:1905.06314},
year = {2019}
}