AIfES: A Next-Generation Edge AI Framework.

Edge Artificial Intelligence (AI) relies on the integration of Machine Learning (ML) into even the smallest embedded devices, thus enabling local intelligence in real-world applications, e.g. for image or speech processing. Traditional Edge AI frameworks lack important aspects required to keep up with recent and upcoming ML innovations. These aspects include low flexibility concerning the target hardware and limited support for custom hardware accelerator integration. Artificial Intelligence for Embedded Systems Framework (AIfES) has the goal to overcome these challenges faced by traditional edge AI frameworks. In this paper, we give a detailed overview of the architecture of AIfES and the applied design principles. Finally, we compare AIfES with TensorFlow Lite for Microcontrollers (TFLM) on an ARM Cortex-M4-based System-on-Chip (SoC) using fully connected neural networks (FCNNs) and convolutional neural networks (CNNs). AIfES outperforms TFLM in both execution time and memory consumption for the FCNNs. Additionally, using AIfES reduces memory consumption by up to 54% when using CNNs. Furthermore, we show the performance of AIfES during the training of FCNN as well as CNN and demonstrate the feasibility of training a CNN on a resource-constrained device with a memory usage of slightly more than 100 kB of RAM.

ComplexEye: a multi-lens array microscope for high-throughput embedded immune cell migration analysis.

Autonomous migration is essential for the function of immune cells such as neutrophils and plays an important role in numerous diseases. The ability to routinely measure or target it would offer a wealth of clinical applications. Video microscopy of live cells is ideal for migration analysis, but cannot be performed at sufficiently high-throughput (HT). Here we introduce ComplexEye, an array microscope with 16 independent aberration-corrected glass lenses spaced at the pitch of a 96-well plate to produce high-resolution movies of migrating cells. With the system, we enable HT migration analysis of immune cells in 96- and 384-well plates with very energy-efficient performance. We demonstrate that the system can measure multiple clinical samples simultaneously. Furthermore, we screen 1000 compounds and identify 17 modifiers of migration in human neutrophils in just 4 days, a task that requires 60-times longer with a conventional video microscope. ComplexEye thus opens the field of phenotypic HT migration screens and enables routine migration analysis for the clinical setting.

Explainable Remaining Tool Life Prediction for Individualized Production Using Automated Machine Learning.

The increasing demand for customized products is a core driver of novel automation concepts in Industry 4.0. For the case of machining complex free-form workpieces, e.g., in die making and mold making, individualized manufacturing is already the industrial practice. The varying process conditions and demanding machining processes lead to a high relevance of machining domain experts and a low degree of manufacturing flow automation. In order to increase the degree of automation, online process monitoring and the prediction of the quality-related remaining cutting tool life is indispensable. However, the varying process conditions complicate this as the correlation between the sensor signals and tool condition is not directly apparent. Furthermore, machine learning (ML) knowledge is limited on the shop floor, preventing a manual adaption of the models to changing conditions. Therefore, this paper introduces a new method for remaining tool life prediction in individualized production using automated machine learning (AutoML). The method enables the incorporation of machining expert knowledge via the model inputs and outputs. It automatically creates end-to-end ML pipelines based on optimized ensembles of regression and forecasting models. An explainability algorithm visualizes the relevance of the model inputs for the decision making. The method is analyzed and compared to a manual state-of-the-art approach for series production in a comprehensive evaluation using a new milling dataset. The dataset represents gradual tool wear under changing workpieces and process parameters. Our AutoML method outperforms the state-of-the-art approach and the evaluation indicates that a transfer of methods designed for series production to variable process conditions is not easily possible. Overall, the new method optimizes individualized production economically and in terms of resources. Machining experts with limited ML knowledge can leverage their domain knowledge to develop, validate and adapt tool life models.

Analytical Evaluation of Signal-to-Noise Ratios for Avalanche- and Single-Photon Avalanche Diodes.

Performance of systems for optical detection depends on the choice of the right detector for the right application. Designers of optical systems for ranging applications can choose from a variety of highly sensitive photodetectors, of which the two most prominent ones are linear mode avalanche photodiodes (LM-APDs or APDs) and Geiger-mode APDs or single-photon avalanche diodes (SPADs). Both achieve high responsivity and fast optical response, while maintaining low noise characteristics, which is crucial in low-light applications such as fluorescence lifetime measurements or high intensity measurements, for example, Light Detection and Ranging (LiDAR), in outdoor scenarios. The signal-to-noise ratio (SNR) of detectors is used as an analytical, scenario-dependent tool to simplify detector choice for optical system designers depending on technologically achievable photodiode parameters. In this article, analytical methods are used to obtain a universal SNR comparison of APDs and SPADs for the first time. Different signal and ambient light power levels are evaluated. The low noise characteristic of a typical SPAD leads to high SNR in scenarios with overall low signal power, but high background illumination can saturate the detector. LM-APDs achieve higher SNR in systems with higher signal and noise power but compromise signals with low power because of the noise characteristic of the diode and its readout electronics. Besides pure differentiation of signal levels without time information, ranging performance in LiDAR with time-dependent signals is discussed for a reference distance of 100 m. This evaluation should support LiDAR system designers in choosing a matching photodiode and allows for further discussion regarding future technological development and multi pixel detector designs in a common framework.

Reproducibility of photoplethysmography-based local pulse transit time measurement.

There are several previously published approaches of measuring local pulse transit time (PTT). One of these approaches is to use two optical sensors based on photoplethysmography (PPG). However, little information about reproducibility in PPG based PTT measurement is available. Therefore, we performed a small sample size study (n = 5) to investigate quantitative criteria for reproducible PTT measurement. The inflection point as a characteristic feature of the pulse wave showed the most stabile results under varying conditions. Furthermore, we found that correlation between related pulse waves could be used as a threshold for signal quality. We suggest to implement a real-time operator feedback based on the found criteria to ensure reproducible PTT measurements.