7 papers indexed
Mahdi Abdollahpour, Marco Bertuletti, Yichao Zhang, Luca Benini, Alessandro Vanelli-Coralli
Attention-based neural estimators achieve strong channel-estimation accuracy, but the computational cost of global attention over the time-frequency resource grid grows quadratically with the number of subcarriers, and these estimators are typically tied to a single resource allo…
Vlad Niculescu, Lorenzo Lamberti, Francesco Conti, Luca Benini, Daniele Palossi
The evolution of energy-efficient ultra-low-power (ULP) parallel processors and the diffusion of convolutional neural networks (CNNs) are fueling the advent of autonomous driving nano-sized unmanned aerial vehicles (UAVs). These sub-10 cm robotic platforms are envisioned as next-…
Sergei Vostrikov, Federico Villani, Cedric Hirschi, Jinhao Lu, Jonas Welsch, Martin Angerer, et al.
Wearable ultrasound enables continuous monitoring of physiological processes such as muscle dynamics, bladder volume, and cardiovascular activity. Existing fully wearable ultra-low-power platforms are limited to shallow, low-channel A-mode sensing, while larger multi-mode systems…
Victor J. B. Jung, Gagandeep Singh, Joseph Melber, Kristof Denolf, Francesco Conti, Luca Benini
The growing adoption of large language model-based agents within operating system workflows has increased the importance of energy-efficient inference on laptop-class systems-on-chip (SoCs). While cloud offloading remains common, it introduces reliability and privacy concerns tha…
Lorenzo Lamberti, Manuele Rusci, Marco Fariselli, Francesco Paci, Luca Benini
In this paper, we present the first (to the best of our knowledge) demonstration of a low-power MCU-based edge device for Automatic License Plate Recognition (ALPR). The design leverages on a 9-core RISC-V processor, GAP8, coupled with a QVGA ultra-low-power greyscale imager. The…
Nicolas Baumann, Liam Boyle, Pu Deng, Edoardo Ghignone, Boyang Sun, Marc Pollefeys, et al.
Vision Language Models (VLMs) have emerged in the robotic domain as a powerful tool that enables environmental perception with language context, serving as a catalyst for open-vocabulary tasks like ObjectNav. Yet, their computational footprint typically confines them to cloud exe…
Thorir Mar Ingolfsson, Wajeeha Tahir, Anna Tegon, Lionnus Kesting, Gamze İslamoğlu, Luca Benini
Recursive reasoning models solve hard puzzles by applying compact, weight-tied blocks over many refinement steps. Because these blocks are reused many times, quantizing them creates a unique dynamical problem: the quantization error is incurred at every step. While 8-bit quantiza…