Machine Learning on Mainstream Microcontrollers

Sakr, Fouad; Bellotti, Francesco; Berta, Riccardo; Gloria, Alessandro De

doi:10.3390/s20092638

Cited by 74 publications

(54 citation statements)

References 27 publications

(22 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the evaluation, we employed an open-source framework [ 56 ] that offers implementations of the ML algorithms (KNN, DT and, SVM) in python (scikit-learn library) and C-code, which we used for training and deployment, respectively. Since the evaluation framework offers the implementations using float32 (FP) data type, we executed them on the NXP k64f MCU, which contains an FP unit.…”

Section: Experimental Setup and Resultsmentioning

confidence: 99%

Robustifying the Deployment of tinyML Models for Autonomous Mini-Vehicles

Prado

Rusci

Capotondi

et al. 2021

Sensors

View full text Add to dashboard Cite

Standard-sized autonomous vehicles have rapidly improved thanks to the breakthroughs of deep learning. However, scaling autonomous driving to mini-vehicles poses several challenges due to their limited on-board storage and computing capabilities. Moreover, autonomous systems lack robustness when deployed in dynamic environments where the underlying distribution is different from the distribution learned during training. To address these challenges, we propose a closed-loop learning flow for autonomous driving mini-vehicles that includes the target deployment environment in-the-loop. We leverage a family of compact and high-throughput tinyCNNs to control the mini-vehicle that learn by imitating a computer vision algorithm, i.e., the expert, in the target environment. Thus, the tinyCNNs, having only access to an on-board fast-rate linear camera, gain robustness to lighting conditions and improve over time. Moreover, we introduce an online predictor that can choose between different tinyCNN models at runtime—trading accuracy and latency—which minimises the inference’s energy consumption by up to 3.2×. Finally, we leverage GAP8, a parallel ultra-low-power RISC-V-based micro-controller unit (MCU), to meet the real-time inference requirements. When running the family of tinyCNNs, our solution running on GAP8 outperforms any other implementation on the STM32L4 and NXP k64f (traditional single-core MCUs), reducing the latency by over 13× and the energy consumption by 92%.

show abstract

Section: Experimental Setup and Resultsmentioning

confidence: 99%

Robustifying the Deployment of tinyML Models for Autonomous Mini-Vehicles

Prado

Rusci

Capotondi

et al. 2021

Sensors

View full text Add to dashboard Cite

show abstract

“…Various operating systems are widely available for the Internet of Things (IoT) environment using embedded devices based on micro-controllers with the characteristics of being low-powered, constrained, and connected [ 64 , 65 ]. At the same time, different environments designed to train ML algorithms in desktop computers to make inferences in micro-controllers have also been tested in the literature [ 66 ]. This paper generalised an environment based on Mbed OS and TensorFlow Lite.…”

Section: Resultsmentioning

confidence: 99%

A Generalization Performance Study Using Deep Learning Networks in Embedded Systems

Gorospe

Mulero²,

Arbelaitz

et al. 2021

Sensors

View full text Add to dashboard Cite

Deep learning techniques are being increasingly used in the scientific community as a consequence of the high computational capacity of current systems and the increase in the amount of data available as a result of the digitalisation of society in general and the industrial world in particular. In addition, the immersion of the field of edge computing, which focuses on integrating artificial intelligence as close as possible to the client, makes it possible to implement systems that act in real time without the need to transfer all of the data to centralised servers. The combination of these two concepts can lead to systems with the capacity to make correct decisions and act based on them immediately and in situ. Despite this, the low capacity of embedded systems greatly hinders this integration, so the possibility of being able to integrate them into a wide range of micro-controllers can be a great advantage. This paper contributes with the generation of an environment based on Mbed OS and TensorFlow Lite to be embedded in any general purpose embedded system, allowing the introduction of deep learning architectures. The experiments herein prove that the proposed system is competitive if compared to other commercial systems.

show abstract

“…The vast majority of these solutions follow a centralized intelligence architecture where there are several devices responsible for taking information from the surrounding environment and acting on it, connected to a controller that contains all the necessary logic to monitor and automate them. Another option would be to integrate the intelligence directly into the actuating devices, enabling them to locally process the collected data, providing faster responses with no need to consume time in communication, however, architectures of this type are still not very common (at least in ML applications) due to the reduced computational capacity of microcontrollers and lack of quantitative analyses about the performance of common ML algorithms on such devices [22].…”

Section: Related Workmentioning

confidence: 99%

Voice-Activated Smart Home Controller Using Machine Learning

2021

View full text Add to dashboard Cite

The emergence of the Internet of Things concept has provided a great vision for the technological future, intending to enable the extraction and comprehension of information from the environment around us, making use of the interaction and cooperation between several technological devices. The example of Smart Homes, in particular, aims to integrate these devices into households, enabling the automation of tasks previously performed by humans, to simplify their daily lives and create a more comfortable environment. However, many of these devices fail to keep their promise, since they were not developed taking into account the frequent change of habits and tastes of the user, being necessary reprogramming of the device to follow the new behaviors. Taking this problem into account, this article presents the design and end-to-end implementation of a voice-activated smart home controller for intelligent devices, deployed in a real environment and validated in an experimental setup of motorized blinds. The architecture of the proposed solution integrates evolvable intelligence with the use of an Online Learning framework, enabling it to automatically adapt to the user's habits and behavioral patterns. The results obtained from the various evaluation tests provide a validation of the operation and usefulness of the developed system. The main contributions of this work are: I) design of a smart home controller's architecture; II) end-to-end implementation of a smart home controller and respective guidelines; III) opensource dataset of user behavior from the smart blinds scenario; IV) comparison between Online and Offline Learning approaches.

show abstract

Machine Learning on Mainstream Microcontrollers

Cited by 74 publications

References 27 publications

Robustifying the Deployment of tinyML Models for Autonomous Mini-Vehicles

Robustifying the Deployment of tinyML Models for Autonomous Mini-Vehicles

A Generalization Performance Study Using Deep Learning Networks in Embedded Systems

Voice-Activated Smart Home Controller Using Machine Learning

Contact Info

Product

Resources

About