Search

This paper deploys and explores variants of TinyissimoYOLO, a highly flexible and fully quantized ultra-lightweight object detection network designed for edge systems with a power envelope of a few milliwatts. With experimental measurements, we present a comprehensive characterization of the network’s detection performance, exploring the impact of various parameters, including input resolution, number of object classes, and hidden layer adjustments. We deploy variants of TinyissimoYOLO on state-of-the-art ultra-low-power extreme edge platforms, presenting a detailed comparison on latency, energy efficiency, and their ability to efficiently parallelize the workload. In particular, the paper presents a comparison between a RISC-V-based parallel processor (GAP9 from GreenWaves Technologies) with and without use of its on-chip hardware accelerator, an ARM Cortex-M7 core (STM32H7 from ST Microelectronics), two ARM Cortex-M4 cores (STM32L4 from ST Microelectronics and Apollo4b from Ambiq), and a multi-core platform aimed at edge AI applications with a CNN hardware accelerator (MAX78000 from Analog Devices). Experimental results show that the GAP9’s hardware accelerator achieves the lowest inference latency and energy at $\mathrm {2.12ms }$ and $\mathrm {150~\mu \text {J} }$ respectively, which is around 2x faster and 20% more energy efficient than the next best platform, the MAX78000. The hardware accelerator of GAP9 can even run an increased resolution version of TinyissimoYOLO with $112\times 112$ pixels and 10 detection classes within 3.2 ms, consuming $\mathrm {245~\mu \text {J} }$ . We also deployed and profiled a multi-core implementation on GAP9 at different core voltages and frequencies, achieving $\mathrm {11.3ms }$ with the lowest-latency and $\mathrm {490~\mu \text {J} }$ with the most energy-efficient configuration. With this paper, we demonstrate the flexibility of TinyissimoYOLO and prove its detection accuracy on a widely used detection dataset. Furthermore, we demonstrate its suitability for real-time ultra-low-power edge inference.

Introduction Professional sports are fiercely competitive. In ski jumping, for example, even small changes in take-off and flight can make a decisive difference between victory and defeat (Elfmark et al., 2022). Within the short time of a jump, athletes must learn to solve complex motor control problems while being exposed to harsh environmental conditions, e.g., wind, snow, and low temperatures. The actual take-off occurs within the blink of an eye (~300 ms) and an aerodynamically favourable and stable flight position should be attained immediately. Fine control of the centre of gravity in the in-run favours high speeds to generate optimum momentum during take-off (Müller, 2008). In flight, athletes can voluntarily influence aerodynamics by changing their body position. However, non-optimal flight positions occur unintentionally or due to incorrect behaviour. Furthermore, as a non-cyclical sport, ski jumping suffers from low repetition rates, which impairs the effectiveness of training. Thus, increasing the learning rate for each jump is a key success factor. Biofeedback methods have been shown to accelerate motor learning in athletes (Mulder & Hulstijn, 1985). Current sensor technologies in ski jumping do not meet the requirements for a truly wearable system, which must be energy-efficient, unobtrusive and barely noticeable (so as not to interfere with natural movement behaviour and jumping technique) and, in particular, must be equipped with a wireless link (for real-time data analysis, e.g. on the trainer tower; Schulthess et al., 2023). Methods The proposed system consists of two multi-sensor nodes: One node is hidden in a modified ski jumping boot, integrating three force-sensing resistor sensors to measure the pressure distribution on the foot soles of ski jumpers. The second sensor node is located in the ski goggles and contains RGB LEDs that provide visual biofeedback in the peripheral vision. Results We have calculated the total power consumption of our systems to be 2.52 mW, meeting requirements for multi-day operation between battery recharges. Our on-device body position classification model achieves an accuracy of 92.7% in recognising body positions from data recorded in the laboratory. Discussion/Conclusion This is the first truly wearable training system in ski jumping, offering professional athletes a new augmented experience, aimed at accelerating motor learning. In addition, the real-time data transmission of biomechanically relevant characteristics facilitates the work of the training team and could in the future enable more informative and entertaining television broadcasts. References Elfmark, O., Ettema, G., & Gilgien, M. (2022). Assessment of the steady glide phase in ski jumping. Journal of Biomechanics, 139, 111139. https://doi.org/10.1016/j.jbiomech.2022.111139 Mulder, T., & Hulstijn, W. (1985). Sensory feedback in the learning of a novel motor task. Journal of Motor Behavior, 17(1), 110–128. https://doi.org/10.1080/00222895.1985.10735340 Müller, W. (2008). Performance factors in ski jumping. In H. Nørstrud (Ed.), Sport Aerodynamics (pp. 139– 160). Springer. https://doi.org/10.1007/978-3-211-89297-8_8 Schulthess, L., Ingolfsson, T. M., Nölke, M., Magno, M., Benini, L., & Leitner, C. (2023). Skilog: A smart sensor system for performance analysis and biofeedback in ski jumping. https://doi.org/10.48550/arXiv.2309.14455

Abstract Current trend in neurosciences is to use naturalistic stimuli, such as cinema, class-room biology or video gaming, aiming to understand the brain functions during ecologically valid conditions. Naturalistic stimuli recruit complex and overlapping cognitive, emotional and sensory brain processes. Brain oscillations form underlying mechanisms for such processes, and further, these processes can be modified by expertise. Human cortical functions are often analyzed with linear methods despite brain as a biological system is highly nonlinear. This study applies a relatively robust nonlinear method, Higuchi fractal dimension (HFD), to classify cortical functions of math experts and novices when they solve long and complex math demonstrations in an EEG laboratory. Brain imaging data, which is collected over a long time span during naturalistic stimuli, enables the application of data-driven analyses. Therefore, we also explore the neural signature of math expertise with machine learning algorithms. There is a need for novel methodologies in analyzing naturalistic data because formulation of theories of the brain functions in the real world based on reductionist and simplified study designs is both challenging and questionable. Data-driven intelligent approaches may be helpful in developing and testing new theories on complex brain functions. Our results clarify the different neural signature, analyzed by HFD, of math experts and novices during complex math and suggest machine learning as a promising data-driven approach to understand the brain processes in expertise and mathematical cognition.

Wind energy generation plays a vital role in transitioning from fossil fuel-based energy sources and in alleviating the impacts of global warming. However, global wind energy coverage still needs to rise, while requiring a significant step up in conversion efficiency: monitoring wind flow and operational parameters of wind turbines is an essential prerequisite for coverage and conversion efficiency optimization. This paper presents a low-power, self-sustainable, and time-synchronised system for aerodynamic and acoustic measurements on operating wind turbines. It includes 40 high-accuracy barometers, 10 microphones, 5 differential pressure sensors, and implements a coarse time synchronisation on top of a Bluetooth Low Energy 5.1 protocol tuned for long-range communications. Moreover, we field-assessed the node capability to collect precise and accurate aerodynamic data with a multi-node setup. Outdoor experimental tests revealed that the system can acquire heterogeneous data with a time synchronisation error below $\mathrm {100~ \mu \text {s} }$ and sustain a data rate of 600 kbps over 400 m with up to 5 sensor nodes, enough to fully instrument a wind turbine. The proposed method does not add any traffic overhead on the Bluetooth Low Energy 5.1 protocol, fully relying only on connection events and withstands transmission discontinuity often present in long range wireless communications.

The awareness of rock shape dependence in rockfall hazard assessment is growing, but experimental and field studies are scarce. This study presents a large data set of induced single block rockfall events quantifying the influence of rock shape and mass on its complex kinematic behaviour.

IPS is a crucial technology that enables medical staff and hospital management to accurately locate and track persons or assets inside medical buildings. Among other technologies, readily available BLE can be exploited to achieve an energy-efficient and low-cost solution. This work presents the design, implementation and comparison of a RSSI-based and a MCPD-based indoor localization system. The implementation is based on a lightweight wkNN algorithm that processes RSSI and MCPD distance data from connection-less BLE Beacons. The designed hardware and firmware are implemented around the state-of-the-art SoC for BLE, the nRF5340 from Nordic Semiconductor. Experimental evaluation with real-time data processing has been evaluated and presented in a 7.3 m × 8.9 m room with furniture and six beacon nodes. The experimental results on randomly chosen validation points within the room show an average error of only 0.50 m for the MCPD approach, whereas the RSSI approach achieved an error of 1.39 m.

Medical devices measure vital parameters such as pulse, respiration rate, and blood oxygenation, over periods of days or weeks in a continuous manner. Traditional systems only support such requirements in stationary applications where a constant power supply is available. Trends toward remote healthcare and telemedicine require wearable devices, able to provide similar functionalities in wireless mode. Miniaturized and thin form factors, desirable in wearable applications, set stringent constraints on the available power, and consequently on the accuracy and lifetime. Energy harvesting combined with low-power design and energy efficient processing can significantly extend the lifetime of wearable devices. This paper presents a wearable pulse oximeter assembled in a 3D ring-like geometry that achieves self-sustainability by exploiting efficient power management, solar energy harvesting, and ultra-low power processing in a multi-core microcontroller. The design strategy of combining onboard processing to monitor blood oxygenation and the transmission of only relevant information via a Bluetooth low-energy (BLE) interface, significantly reduces the overall energy consumption. Experimental results on the designed and developed prototype demonstrate that measuring the blood oxygenation once every minute with a sampling rate of 100 samples/s achieve accurate results at the daily energy consumption of 28 J including hourly BLE transmissions. The low-power design allows the system to be self-sustainable with just 64 min of sunlight per day or 12 hrs. of indoor home light.

Low-power energy harvesting has been demonstrated as a feasible alternative for the power supply of next-generation smart sensors and IoT end devices. In many cases, the output of kinetic energy harvesters is an alternating current (AC) requiring rectification in order to supply the electronic load. The rectifier design and selection can have a considerable influence on the energy harvesting system performance in terms of extracted output power and conversion losses. This paper presents a quantitative comparison of three passive rectifiers in a low-power, low-voltage electromagnetic energy harvesting sub-system, namely the full-wave bridge rectifier (FWR), the voltage doubler (VD), and the negative voltage converter rectifier (NVC). Based on a variable reluctance energy harvesting system, we investigate each of the rectifiers with respect to their performance and their effect on the overall energy extraction. We conduct experiments under the conditions of a low-speed rotational energy harvesting application with rotational speeds of 5 rpm to 20 rpm, and verify the experiments in an end-to-end energy harvesting evaluation. Two performance metrics—power conversion efficiency (PCE) and power extraction efficiency (PEE)—are obtained from the measurements to evaluate the performance of the system implementation adopting each of the rectifiers. The results show that the FWR with PEEs of 20% at 5 rpm to 40% at 20 rpm has a low performance in comparison to the VD (40–60%) and NVC (20–70%) rectifiers. The VD-based interface circuit demonstrates the best performance under low rotational speeds, whereas the NVC outperforms the VD at higher speeds (>18 rpm). Finally, the end-to-end system evaluation is conducted with a self-powered rpm sensing system, which demonstrates an improved performance with the VD rectifier implementation reaching the system’s maximum sampling rate (40 Hz) at a rotational speed of approximately 15.5 rpm.

Optogenetics provide a potential alternative approach to the treatment of chronic pain, in which complex pathology often hampers efficacy of standard pharmacological approaches. Technological advancements in the development of thin, wireless, and mechanically flexible optoelectronic implants offer new routes to control the activity of subsets of neurons and nerve fibers in vivo. This study reports a novel and advanced design of battery-free, flexible, and lightweight devices equipped with one or two miniaturized LEDs, which can be individually controlled in real time. Two proof-of-concept experiments in mice demonstrate the feasibility of these devices. First, we show that blue-light devices implanted on top of the lumbar spinal cord can excite channelrhodopsin expressing nociceptors to induce place aversion. Second, we show that nocifensive withdrawal responses can be suppressed by green-light optogenetic (Archaerhodopsin-mediated) inhibition of action potential propagation along the sciatic nerve. One salient feature of these devices is that they can be operated via modern tablets and smartphones without bulky and complex lab instrumentation. In addition to the optical stimulation, the design enables the simultaneously wireless recording of the temperature in proximity of the stimulation area. As such, these devices are primed for translation to human patients with implications in the treatment of neurological and psychiatric conditions far beyond chronic pain syndromes.

LoRa is popular for internet of things applications as this communication technology offers both a long range and a low power consumption. However, LoRaWAN, the standard MAC protocol that uses LoRa as physical layer, has the bottleneck of a high downlink latency to achieve energy efficiency. To overcome this drawback we explore the use of wake-up radio combined with LoRa, and propose an adequate MAC protocol that takes profit of both these heterogeneous and complementary technologies. This protocol allows an opportunistic selection of a cluster head that forwards commands from the gateway to the nodes in the same cluster. Furthermore, to achieve self-sustainability, sensor nodes might include an energy harvesting sub-system, for instance to scavenge energy from the light, and their quality of service can be tuned, according to their available energy. To have an effective self-sustaining LoRa system, we propose a new energy manager that allows less fluctuations of the quality of service between days and nights. Latency and energy are modeled in a hybrid manner, i.e., leveraging microbenchmarks on real hardware platforms, to explore the influence of the energy harvesting conditions on the quality of service of this heterogeneous network. It is clearly demonstrated that the cooperation of nodes within a cluster drastically reduces the latency of LoRa base station commands, e.g., by almost 90% compared to traditional LoRa scheme for a 10 nodes cluster.

We report on a self-sustainable, wireless accelerometer-based system for wear detection in a band saw blade. Due to the combination of low power hardware design, thermal energy harvesting with a small thermoelectric generator (TEG), an ultra-low power wake-up radio, power management and the low complexity algorithm implemented, our solution works perpetually while also achieving high accuracy. The onboard algorithm processes sensor data, extracts features, performs the classification needed for the blade’s wear detection, and sends the report wirelessly. Experimental results in a real-world deployment scenario demonstrate that its accuracy is comparable to state-of-the-art algorithms executed on a PC and show the energy-neutrality of the solution using a small thermoelectric generator to harvest energy. The impact of various low-power techniques implemented on the node is analyzed, highlighting the benefits of onboard processing, the nano-power wake-up radio, and the combination of harvesting and low power design. Finally, accurate in-field energy intake measurements, coupled with simulations, demonstrate that the proposed approach is energy autonomous and can work perpetually.

Energy efficiency is crucial in the design of battery-powered end devices, such as smart sensors for the Internet of Things applications. Wireless communication between these distributed smart devices consumes significant energy, and even more when data need to reach several kilometers in distance. Low-power and long-range communication technologies such as LoRaWAN are becoming popular in IoT applications. However, LoRaWAN has drawbacks in terms of (i) data latency; (ii) limited control over the end devices by the gateway; and (iii) high rate of packet collisions in a dense network. To overcome these drawbacks, we present an energy-efficient network architecture and a high-efficiency on-demand time-division multiple access (TDMA) communication protocol for IoT improving both the energy efficiency and the latency of standard LoRa networks. We combine the capabilities of short-range wake-up radios to achieve ultra-low power states and asynchronous communication together with the long-range connectivity of LoRa. The proposed approach still works with the standard LoRa protocol, but improves performance with an on-demand TDMA. Thanks to the proposed network and protocol, we achieve a packet delivery ratio of 100% by eliminating the possibility of packet collisions. The network also achieves a round-trip latency on the order of milliseconds with sensing devices dissipating less than 46 mJ when active and 1.83 μ W during periods of inactivity and can last up to three years on a 1200-mAh lithium polymer battery.

Wireless sensor nodes have a limited power budget, though they are often expected to be functional in the field once deployed for extended periods of time. Therefore, minimization of energy consumption and energy harvesting technology in Wireless Sensor Networks (WSN) are key tools for maximizing network lifetime, and achieving self-sustainability. This paper proposes an energy aware Adaptive Sampling Algorithm (ASA) for WSN with power hungry sensors and harvesting capabilities, an energy management technique that can be implemented on any WSN platform with enough processing power to execute the proposed algorithm. An existing state-of-the-art ASA developed for wireless sensor networks with power hungry sensors is optimized and enhanced to adapt the sampling frequency according to the available energy of the node. The proposed algorithm is evaluated using two in-field testbeds that are supplied by two different energy harvesting sources (solar and wind). Simulation and comparison between the state-of-the-art ASA and the proposed energy aware ASA (EASA) in terms of energy durability are carried out using in-field measured harvested energy (using both wind and solar sources) and power hungry sensors (ultrasonic wind sensor and gas sensors). The simulation results demonstrate that using ASA in combination with an energy aware function on the nodes can drastically increase the lifetime of a WSN node and enable self-sustainability. In fact, the proposed EASA in conjunction with energy harvesting capability can lead towards perpetual WSN operation and significantly outperform the state-of-the-art ASA.