The lack of attention during the driving task is considered as a major risk factor for fatal road accidents around the world. Despite the ever-growing trend for autonomous driving which promises to bring greater road-safety benefits, the fact is today’s vehicles still only feature partial and conditional automation, demanding frequent driver action. Moreover, the monotony of such a scenario may induce fatigue or distraction, reducing driver awareness and impairing the regain of the vehicle’s control. To address this challenge, we introduce a non-intrusive system to monitor the driver in terms of fatigue, distraction, and activity. The proposed system explores state-of-the-art sensors, as well as machine learning algorithms for data extraction and modeling. In the domain of fatigue supervision, we propose a feature set that considers the vehicle’s automation level. In terms of distraction assessment, the contributions concern (i) a holistic system that covers the full range of driver distraction types and (ii) a monitoring unit that predicts the driver activity causing the faulty behavior. By comparing the performance of Support Vector Machines against Decision Trees, conducted experiments indicated that our system can predict the driver’s state with an accuracy ranging from 89% to 93%.

Undeniably, the Internet of Things (IoT) ecosystem continues to evolve at a breakneck pace, exceeding all growth expectations and ubiquity barriers. From sensor to cloud, this giant network keeps breaking technological bounds in several domains, and wireless sensor nodes (motes) are expected to be predominant as the number of IoT devices grows towards the trillions. However, their future in the IoT ecosystem still seems foggy, where several challenges, such as (i) device’s connectivity, (ii) intelligence at the edge, (iii) security and privacy concerns, and (iv) growing energy needs, keep pulling in opposite directions. This prospective paper offers a succinct and forward-looking review of recent trends, challenges, and state-of-the-art solutions of low-end IoT motes, where reconfigurable computing technology plays a key role in tomorrow’s IoT devices.

The end of Moore’s Law aligned with data privacy concerns is forcing machine learning (ML) to shift from the cloud to the deep edge. In the next-generation ML systems, the inference and part of the training process will perform at the edge, while the cloud stays responsible for major updates. This new computing paradigm, called federated learning (FL), alleviates the cloud and network infrastructure while increasing data privacy. Recent advances empowered the inference pass of quantized artificial neural networks (ANNs) on Arm Cortex-M and RISC-V microcontroller units (MCUs). Nevertheless, the training remains confined to the cloud, imposing the transaction of high volumes of private data over a network and leading to unpredictable delays when ML applications attempt to adapt to adversarial environments. To fill this gap, we make the first attempt to evaluate the feasibility of ANN training in Arm Cortex-M MCUs. From the available optimization algorithms, stochastic gradient descent (SGD) has the best trade-off between accuracy, memory footprint, and latency. However, its original form and the variants available in the literature still do not fit the stringent requirements of Arm Cortex-M MCUs. We propose L-SGD, a lightweight implementation of SGD optimized for maximum speed and minimal memory footprint in this class of MCUs. We developed a floating-point version and another that operates over quantized weights. For a fully-connected ANN trained on the MNIST dataset, L-SGD (float-32) is 4.20× faster than the SGD while requiring only 2.80% of the memory with negligible accuracy loss. Results also show that quantized training is still unfeasible to train an ANN from the scratch but is a lightweight solution to perform minor model fixes and counteract the fairness problem in typical FL systems.

Capsule networks (CapsNets) are an emerging trend in image processing. In contrast to a convolutional neural network, CapsNets are not vulnerable to object deformation, as the relative spatial information of the objects is preserved across the network. However, their complexity is mainly related to the capsule structure and the dynamic routing mechanism, which makes it almost unreasonable to deploy a CapsNet, in its original form, in a resource-constrained device powered by a small microcontroller (MCU). In an era where intelligence is rapidly shifting from the cloud to the edge, this high complexity imposes serious challenges to the adoption of CapsNets at the very edge. To tackle this issue, we present an API for the execution of quantized CapsNets in Arm Cortex-M and RISC-V MCUs. Our software kernels extend the Arm CMSIS-NN and RISC-V PULP-NN to support capsule operations with 8-bit integers as operands. Along with it, we propose a framework to perform post-training quantization of a CapsNet. Results show a reduction in memory footprint of almost 75%, with accuracy loss ranging from 0.07% to 0.18%. In terms of throughput, our Arm Cortex-M API enables the execution of primary capsule and capsule layers with medium-sized kernels in just 119.94 and 90.60 ms, respectively (STM32H755ZIT6U, Cortex-M7 @ 480 MHz). For the GAP-8 SoC (RISC-V RV32IMCXpulp @ 170 MHz), the latency drops to 7.02 and 38.03 ms, respectively.

8 k ultra-high definition (UHD) is paving the way for the next-generation video systems. In the audiovisual industry, besides delivering a more immersive experience, it is a mean to smooth spatial artifacts during video sampling. In the medical industry, it may provide surgeons with increased reality in surgeries such as endoscopy. Nevertheless, researchers are struggling to meet the high throughput required by this resolution and hardware solutions miss the flexibility required for on-demand updates. In this context, we propose an 8 k video-card based on a hybrid platform endowed with "soft" programmable logic and "hard" processors. The former accelerates the video pipeline from capture to encoding and playback-via SDI or PCIe. The later connects with a cloud to stream video and interfaces with the user. Results showed that our platform outputs 8 k UHD video in YUV 4:2:2 10-bits and H.264 formats at 60 and 30 frames per second, respectively.