Search

The large amount of memory usage in recent machine learning applications imposes a significant system burden with respect to power and processing speed. To cope with such a problem, Processing-In-Memory (PIM) techniques can be applied as an alternative solution. Especially, the recommendation system, which is one of the major machine learning applications used in data centers, requires a large memory capacity and therefore represents a suitable candidate application that could be helped by the PIM technique. In this paper, we introduce a machine learning framework, PIMCaffe, designed for in-memory neural processing units and its evaluation environment. PIMCaffe consists of two components: a Caffe2-based deep learning framework that supports PIM acceleration and a PIM-emulating hardware platform. We develop a suite of functions, libraries, application programming interfaces, and a device driver to support the framework. In addition, we implement a prototype Neural Processing Unit (NPU) in PIMCaffe to evaluate the performance of our platform with machine learning applications. Our prototype NPU design includes a vector processor for parallel vector processing and a systolic array unit for matrix multiplication. Using the proposed software framework, we perform a detailed analysis of the in-memory neural processing unit. PIMCaffe supports evaluations of recommendation systems and various convolutional neural network models on the in-memory neural processing unit. PIMCaffe with the NPU shows up to $2.26\times $ , $5.99\times $ , and $1.71\times $ speedup, compared to the ARM Cortex-A53 CPU, for the recommendation system, AlexNet, and ResNet-50, respectively.

The modern distribution automation suggests to enable decentralized self-healing of distribution networks using advance metering and controlling infrastructure. Decentralized fault location and isolation, as an essential and vital component of self-healing, has attracted considerable attention over the years. In this paper, an integrated fault location and isolation strategy based on fully decentralized multi-agents system (FDMAS) is presented for distribution system with distributed generators (DGs) by combining the primary protection with device failure protection (i.e., backup protection). The proposed FDMAS strategy can locate and isolate electrical fault rapidly even under device failures with minimum fault clearance time and range by using expert logical rules, meanwhile can detect and identify device failure adaptively. Furthermore, a unified programming framework is developed for generalization and application of the proposed strategy. The simulation studies are carried out on 22-bus distribution system by using dynamic model test platform. The test results show that the proposed strategy has an excellent performance on fault clearance time, fault isolation range and device failure detection.

As quantum internet technologies develop, the need for simulation software and education for quantum internet rises. QuNetSim aims to fill this need. QuNetSim is a Python software framework that can be used to simulate quantum networks up to the network layer. The goal of QuNetSim is to make it easier to investigate and test quantum networking protocols over various quantum network configurations and parameters. The framework incorporates many known quantum network protocols so that users can quickly build simulations and beginners can easily learn to implement their own quantum networking protocols., 11 pages, 6 figures. To appear in IEEE Transactions on Quantum Engineering

The Hadoop platform provides a powerful software framework for distributed storage and processing of massive amounts of data. It is at the heart of big data processing and has found numerous applications in diverse areas, ranging from environmental monitoring to security analysis. To facilitate the storage and processing of big data, a Hadoop platform typically runs on a cluster of servers and may scale up to process big data over thousands of hardware nodes. However, the growing scale and complexity of the Hadoop platform also make it increasingly challenging to manage and operate. In this paper, we present a framework called LogM that leverages not only the deep learning model, but also the knowledge graph technology for failure prediction and analysis of the Hadoop cluster. In particular, we first develop a CAB net (Convolutional Neural Network (CNN) with attention-based Bi-directional Long Short Term Memory (Bi-LSTM)) architecture to effectively learn the temporal dynamics from the sequential log data, which allows us to predict system failures. We then adopt a knowledge graph approach for failure analysis and diagnosis. Extensive experiments have been carried out to assess the performance of the proposed approach. It is seen that LogM is highly effective in predicting and diagnosing system failures.

We introduce a software framework for design and programming of large scale programmable photonic chips. The framework facilitates design, simulation, configuration and measurements of reconfigurable waveguide meshes, managing hundreds of actuators and translate high-level user requirements into driving strategies of the electronic channels.

An accurate and efficient intelligent fault diagnosis of mobile robotic roller bearings can significantly enhance the reliability and safety of mechanical systems. To improve the efficiency of intelligent fault classification of mobile robotic roller bearings, this paper proposes a parallel machine learning algorithm using fine-grained-mode Spark on a Mesos big data cloud computing software framework. Through the segmentation of datasets and the support of a parallel framework, the parallel processing technology Spark is combined with a support vector machine (SVM), and a parallel single-SVM algorithm is realized using Scala language. In this approach, empirical mode decomposition (EMD) is applied to extract the energy of the acceleration vibration signal in different frequency bands as features. The parallel EMD-SVM approach is applied to detect faults in mobile robotic roller bearings from fault vibration signals. The experimental results show that it can accurately and effectively identify the faults, and it outperforms existing methods based on parallel deep belief network (DBN) and parallel radial basis function neural network under different training set sizes. Fault classification tests are conducted on outer-race and inner-race faults: in both cases, the proposed parallel EMD-SVM outperforms the serial EMD-SVM in terms of the classification accuracy and classification time under different test sizes. For a small number of nodes, the processing time of the proposed Spark model is less than that of Hadoop but close to that of Storm. For 17 slave nodes, the average precision, average recall, and average F1 score of Spark on Mesos in the fine-grained mode reach 97.3, 97.8, and 97.9%, respectively. The parallel EMD-SVM algorithm in the big data Spark cloud computing framework can improve the accuracy of intelligent fault classification, albeit by a small margin, with higher processing speed and learning convergence rate.

Commodity operating systems are considered vulnerable. Therefore, when an application handles security-sensitive data, it is highly recommended to run the application in a trusted execution environment. In response to this demand, hardware-based trusted execution environments such as Intel SGX and ARM TrustZone have been developed in commodity computers. However, hardware-based approaches cannot be quickly upgraded to address design vulnerabilities or to reflect customer feedback. In this paper, we propose SofTEE, a software framework to support a trusted execution environment for user applications. For a trusted execution environment, SofTEE should support memory isolation and attestation. For memory isolation, SofTEE relies on kernel deprivileging which delegates the execution of privileged operations such as memory management, from a kernel to a special module called a security monitor. To reduce the overhead of switching between the deprivileged kernel and the security monitor, SofTEE proposes an efficient management mechanism of the address space identifier. SofTEE supports attestation by assuming minimal hardware functionalities of random entropy and root of trust. The main challenge of SofTEE is to guarantee security properties like confidentiality and integrity of security-sensitive applications. For security analysis, we have identified security invariants that SofTEE should meet for confidentiality and integrity guarantees. Based on the security invariants, we have designed and prototyped each component of SofTEE on a Raspberry Pi 3 board. SofTEE produces about 3% overhead in case of a security-sensitive application with long execution time and 23% overhead in case of a security-sensitive application with short execution time.

Graphical PCB design tools like KiCAD lack support for high-level abstraction such as functions and loops. To improve PCB design productivity, we hereby present BHDL, a programming framework for PCB designs. In its compact and declarative syntax, schematics and layouts can be modeled effectively and expressed concisely. Treating all circuits, even a resistor, as functions, BHDL naturally supports modularized development that builds a complex design up from smaller designs hierarchically. As an embedded Domain Specific Language (eDSL), BHDL allows users to leverage the full feature of the host language for customization and extension. Our Jupyter kernel supports web-based, REPL-style development and generates auto-placed PCBs.

Bodily signals can complement subjective and behavioral measures to analyze human factors, such as user engagement or stress, when interacting with virtual reality (VR) environments. Enabling widespread use of (also the real-time analysis) of bodily signals in VR applications could be a powerful method to design more user-centric, personalized VR experiences. However, technical and scientific challenges (e.g., cost of research-grade sensing devices, required coding skills, expert knowledge needed to interpret the data) complicate the integration of bodily data in existing interactive applications. This paper presents the design, development, and evaluation of an open-source software framework named Excite-O-Meter. It allows existing VR applications to integrate, record, analyze, and visualize bodily signals from wearable sensors, with the example of cardiac activity (heart rate and its variability) from the chest strap Polar H10. Survey responses from 58 potential users determined the design requirements for the framework. Two tests evaluated the framework and setup in terms of data acquisition/analysis and data quality. Finally, we present an example experiment that shows how our tool can be an easy-to-use and scientifically validated tool for researchers, hobbyists, or game designers to integrate bodily signals in VR applications.

Realistic modelling of qubit systems including noise and constraints imposed by control hardware is required for performance prediction and control optimization of quantum processors. We introduce qopt, a software framework for simulating qubit dynamics and robust quantum optimal control considering common experimental situations. To this end, we model open and closed qubit systems with a focus on the simulation of realistic noise characteristics and experimental constraints. Specifically, the influence of noise can be calculated using Monte Carlo methods, effective master equations or with the efficient filter function formalism, which enables the investigation and mitigation of auto-correlated noise. In addition, limitations of control electronics including finite bandwidth effects can be considered. The calculation of gradients based on analytic results is implemented to facilitate the efficient optimization of control pulses. The software is published under an open source license, well-tested and features a detailed documentation.

Quantum computing is a technology that promises to offer significant advantages during the coming decades. Though the technology is still in a prototype stage, the last few years have seen many of these prototype devices become accessible to the public. This has been accompanied by the open-source development of the software required to use and test quantum hardware in increasingly sophisticated ways. Such tools provide new education opportunities, not just for quantum computing specifically, but also more broadly for quantum information science and even quantum physics as a whole. In this paper we present a case study of one education resource which aims to take advantage of the opportunities: the open-source online textbook `Learn Quantum Computation using Qiskit'. An overview of the topics covered is given, as well as an explanation of the approach taken for each.

Visual augmented reality (AR) headsets have the potential to enhance surgical navigation by providing physicians with an egocentric visualization interface capable of seamlessly blending the virtual navigation aid with the real surgical scenario. However, technological and human-factor limitations still hinder the routine use of commercial AR headsets in clinical practice. The aim of this work is to unveil the AR rendering pipeline of a device-agnostic software framework conceived to fulfill strict requirements towards the realization of a functional and reliable AR-based surgical navigator and capable of supporting the deployment of AR applications for image-guided surgery on different AR headsets. The AR rendering pipeline provides highly accurate AR overlay under both video and optical see-through modalities with almost no perceivable difference in terms of perception of relative distances and depths when used in the peripersonal space. The rendering pipeline allows the setting of the intrinsic and extrinsic projection parameters of the virtual rendering cameras offline and at runtime: under video see-through modality, the rendering pipeline can be modified to adapt the warping of the camera frames and pursue an orthostereoscopic and almost natural perception of the real scene in the peripersonal space. Similarly, under optical see-through modality, the calibrated intrinsic and extrinsic parameters of the eye-display model can be updated by the user to account for the actual user’s eye position. The results of the performance tests with an eye-replacement camera show an average motion-to-photon latency of around 110 ms for both AR rendering modalities. The AR platform for surgical navigation has already proven its efficacy and reliability under VST modality during real surgical operations in craniomaxillofacial surgery.

Ptolemy II [1] is an open-source software framework for modelling, simulation and design of concurrent, heterogeneous, real-time systems, including distributed/parallel systems [2]. These systems can be described using the combination of different formal as well as computation models. Ptolemy also includes machine learning libraries providing support for particle filtering, model-predictive control, hidden Markov models, and various statistical analysis tools. However, one of the main problems Ptolemy users face is the lack of fundamental recurrent neural network structures. In this paper, an LSTM and a GRU recurrent neural network are implemented in Ptolemy II framework, in order to extend its machine learning library, enabling users to develop their complex recurrent neural networks in significantly less time. The presented work has been verified through a real-world weather forecasting use case; the results demonstrate that our approach has identical accuracy with one of the most widely used machine learning library (i.e. Keras) in all cases. To further increase the impact of our approach, the complete source code is freely distributed to the community.

There are an increasing number of Artificial Intelligence (AI) and Internet of Things (IoT) devices and applications which are used across the globe in different areas. One such area is IoT-enabled smart buildings. There is a growing interest among government and non-government organizations in IoT-enabled smart buildings. Despite the perceived benefits of being smart, there is also a key concern of human safety in an emergency situation. This research aims to address this important concern and will develop an initial framework, called here the Safe'Tech, for human safety in IoT enabled smart buildings. The SafeTech will be developed and evaluated using the well-known design science research method. The SafeTech will comprise of both a design and software prototype solution, which can be used to ensure the safety of humans during an emergency through the collection, processing and communication of real-time data from the IoT sensors embedded in the smart building such as smart campus and shopping centre. The SafeTech will interact with the smart building through sensors and collect the state of the physical environment and sense any emergency and then also interact with the humans in the smart building and inform and direct them about the emergency and guide them for the safe evacuation. The applicability and evaluation of the framework software prototype will be demonstrated with the help of smart building scenarios in both Australia and Qatar, involving the fire emergency exercise and automatic and directed evacuation using the SafeTech. This research has several implications, in particular, the impact of reducing human life loss or injury reduction.

Autonomously navigating a robot in an unknown scene comprises tasks such as mapping and localization from the percepts via sensors, safe motion-planning under partial knowledge of the scene, and finally, plan execution using the available motion primitives of the robot. Though many software systems are available for these sub-tasks, there is a lack of integrated software that support autonomous navigation. In this paper, we present an integrated software framework comprising of simultaneous localization and mapping module using visual odometry, a motion-planning module based on constraint-solving, and a plan-execution module particularly designed for a lizard-inspired quadruped robot. To demonstrate its efficacy, we present a number of navigation tasks to the software framework in various indoor simulation settings together with a model of the lizard-inspired quadruped robot and show the results.

DC- DC converters control the power flow and thus the power distribution between the components on different voltage levels. They are essential for (dis)charging batteries and influence the safety and stability of the entire battery management system (BMS). Therefore, testing the functionality and the reliability of DC-DC converters is crucial. This is especially true for decentralized battery management systems (DBMS), where multiple nodes communicate to collectively control the system. The used DC-DC converters are modified to parameterize them during operation via microcontroller interfaces. Integrating the communication into the control loop requires an analysis of the control behavior due to additional delays. Therefore, this paper proposes a framework to test DC-DC converters considering the control and communication perspectives. The response time, the control accuracy and stability of these DC-DC converters, e.g., under continuous and abrupt load changes, are measured in automated tests. The dedicated software framework simulates the DBMS and stimulates the hardware components (e.g. electronic loads, data acquisition) via respective interfaces (CAN, RS232). This allows the test of various DC-DC converters with flexibly adaptable load and power generation profiles. An initial application validates the test framework by verifying the aforementioned aspects and thus the applicability of a DC-DC converter within the DBMS.

In the past few years, microservice architecture has become a dominant software development and deployment paradigm and has aroused interest from both academia and industry. Although the microservice architecture has many advantages, it still faces some challenges. Due to the mobility and changing needs of users, microservice systems (MSS) need to continuously evolve to ensure that the Quality of Service (QoS) perceived by users remains at a satisfactory level. Besides, because of the complexity of MSS constituted by plenty of microservices and users, it is difficult for developers to evolve the MSS manually, and automatic evolutionary adaptability is necessary. However, existing technologies do not well support the automatically adaptive evolution of MSS. To meet this challenge, this paper summarizes common abilities that are necessary for automatically adaptive evolution in MSS, including monitor system runtime status, generate an evolution plan and execute an evolution plan, and an Evolution-Oriented Programming Framework for Microservices (EPF4M) is designed to support individual microservices to achieve those abilities. Considering both the development cost and operational efficiency of microservices, EPF4M is implemented by lightweight methods in Java, including annotations. Experiments are carried out to verify the usability of the functions provided by EPF4M and compare an MSS before and after using EPF4M to demonstrate the acceptability of the additional cost of EPF4M for MSS development and operation. An example is also presented to illustrate the performance of EPF4M in adaptive evolution.

In this work a new and open source software framework for vein recognition is presented, which supports a complete tool chain to conduct biometric experiments. The software can be used out of the box on several publicly available databases and offers a convenient mechanism to configure the tool chain to fit the needs for individual experimental setups. It is implemented in C++ and works on Linux as well as on Windows and therefore offers an advantage over existing vein recognition libraries. The project’s aim is to offer a valuable tool to conduct research in the field of vein recognition and importantly, supporting reproducible research.

Workflow management systems (WMSs) are commonly used to organize/automate sequences of tasks as workflows to accelerate scientific discoveries. During complex workflow modeling, a local interactive workflow environment is desirable, as users usually rely on their rich, local environments for fast prototyping and refinements before they consider using more powerful computing resources. However, existing WMSs do not simultaneously support local interactive workflow environments and HPC resources. In this paper, we present an on-demand access mechanism to remote HPC resources from desktop/laptop-based workflow management software to compose, monitor and analyze scientific workflows in the CyberWater project. Cyber-Water is an open-data and open-modeling software framework for environmental and water communities. In this work, we extend the open-model, open-data design of CyberWater with on-demand HPC accessing capacity. In particular, we design and implement the LaunchAgent library, which can be integrated into the local desktop environment to allow on-demand usage of remote resources for hydrology-related workflows. LaunchAgent manages authentication to remote resources, prepares the computationally-intensive or data-intensive tasks as batch jobs, submits jobs to remote resources, and monitors the quality of services for the users. LaunchAgent interacts seamlessly with other existing components in CyberWater, which is now able to provide advantages of both feature-rich desktop software experience and increased computation power through on-demand HPC/Cloud usage. In our evaluations, we demonstrate how a hydrology workflow that consists of both local and remote tasks can be constructed and show that the added on-demand HPC/Cloud usage helps speeding up hydrology workflows while allowing intuitive workflow configurations and execution using a desktop graphical user interface.

This paper presents development of a Centralized Management System (CMS) software that helps in facilitating smooth charging of Electric Vehicles (EV). CMS encompasses a web application, developed using PHP and HTML; connected to the database, implemented in MySQL, through a local server and also a mobile application, developed in Android Studio. The communication abilities of Electric Vehicle Supply Equipment (EVSE) and EV are realized using Arduino, ESP8266 Wi-Fi modem and CAN modules. The comprehensive communication between EV and EVSE is executed over CAN/Wi-Fi as part of the charging system and EVSE - server data exchange transpires over an Ethernet-based network. This work showcases a software framework and its proficiencies that are needed for a smooth EV charging. The work will help in comprehending the needs and capabilities of a full-fledged communication link and software facilities to realize smooth and flexible EV charging.

As many robot automation applications increasingly rely on multi-core processing or deep-learning models, cloud computing is becoming an attractive and economically viable resource for systems that do not contain high computing power onboard. Despite its immense computing capacity, it is often underused by the robotics and automation community due to lack of expertise in cloud computing and cloud-based infrastructure. Fog Robotics balances computing and data between cloud edge devices. We propose a software framework, FogROS, as an extension of the Robot Operating System (ROS), the defacto standard for creating robot automation applications and components. It allows researchers to deploy components of their software to the cloud with minimal effort, and correspondingly gain access to additional computing cores, GPUs, FPGAs, and TPUs, as well as predeployed software made available by other researchers. FogROS allows a researcher to specify which components of their software will be deployed to the cloud and to what type of computing hardware. We evaluate FogROS on 3 examples: (1) simultaneous localization and mapping (ORB-SLAM2), (2) Dexterity Network (Dex-Net) GPU-based grasp planning, and (3) multi-core motion planning using a 96-core cloud-based server. In all three examples, a component is deployed to the cloud and accelerated with a small change in system launch configuration, while incurring additional latency of 1.2 s, 0.6 s, and 0.5 s due to network communication, the computation speed is improved by 2.6×, 6.0× and 34.2×, respectively. Code, videos, and supplementary material can be found at https://github.com/BerkeleyAutomation/FogROS.

Technological advancements in creating and commercializing novel unobtrusive and wearable physiological sensors generate new opportunities to develop adaptive human-robot interaction (HRI) scenarios. Detecting complex human states such as engagement and stress when interacting with social agents could bring numerous advantages to create meaningful interactive experiences. Despite being widely used to explain human behaviors in post-interaction analysis with social agents, using bodily signals to create more adaptive and responsive systems remains an open challenge. This paper presents the development of an open-source, integrative, and modular library created to facilitate the design of physiologically adaptive HRI scenarios. The HRI Physio Lib streamlines the acquisition, analysis, and translation of human body signals to additional dimensions of perception in HRI applications using social robots. The software framework has four main components: signal acquisition, processing and analysis, social robot and communication, and scenario and adaptation. Information gathered from the sensors is synchronized and processed to allow designers to create adaptive systems that can respond to detected human states. This paper describes the library and presents a use case that uses a humanoid robot as a cardio-aware exercise coach that uses heartbeats to adapt the exercise intensity to maximize cardiovascular performance. The main challenges, lessons learned, scalability of the library, and implications of the physio-adaptive coach are discussed.

With the development of computer network technology, remote working condition monitoring system is gradually applied to the fields of traffic and road condition, construction site, and oil well site monitoring. The traditional monitoring system can achieve its basic functions to a certain extent, but on the one hand, it does not solve the problem of simultaneous transmission of working condition data and realtime images; on the other hand, the networking is not flexible enough. Based on the embedded development board and wireless bridge communication technology, this paper constructs an embedded remote working condition monitoring system. This system combines the V4L2 (Video for linux2) programming framework provided by Linux, and transplants the collection and sending program of working condition data and image data in the development board. First, the development board collects and packs the data; then, the data is transmitted to the upper monitoring center through the wireless bridge to realize real-time monitoring of the scene and real-time display and alarm of working condition data. This paper designs and implements an embedded remote working condition monitoring system, and builds an experimental platform to verify its feasibility.

With the growing interest in Internet of Things (IoT) devices, a number of communication protocols have been developed to support a variety of IoT use cases. One promising communication paradigm that has been widely adopted in the IoT is the publish-subscribe pattern, which is supported by a number of messaging protocols such as MQTT, AMQP, and XMPP. Due to the diversity of IoT device types, an IoT application may communicate with IoT devices using a variety of messaging protocols, software frameworks, and strategies. To this extent, it becomes critical to determine the robustness of components responsible for message delivery (i.e., message brokers). We conduct a comparative study of the MQTT protocol's performance in this paper, comparing performance variables across a range of payload sizes and security levels. Preliminary results indicate that when the payload size remains small, using higher security levels does not result in significant latency overheads. Additionally, we discovered that implementing mutual authentication via Transport Layer Security (TLS) has no effect on MQTT response times in persistent connections when compared to using the default security level, which authenticates only the server.

This paper describes the hardware, software framework, and experimental testing of SURENA IV humanoid robotics platform. SURENA IV has 43 degrees of freedom (DoFs), including seven DoFs for each arm, six DoFs for each hand, and six DoFs for each leg, with a height of 170 cm and a mass of 68 kg and morphological and mass properties similar to an average adult human. SURENA IV aims to realize a cost-effective and anthropomorphic humanoid robot for real-world scenarios. In this way, we demonstrate a locomotion framework based on a novel and inexpensive predictive foot sensor that enables walking with 7cm foot position error because of accumulative error of links and connections’ deflection(that has been manufactured by the tools which are available in the Universities). Thanks to this sensor, the robot can walk on unknown obstacles without any force feedback, by online adaptation of foot height and orientation. Moreover, the arm and hand of the robot have been designed to grasp the objects with different stiffness and geometries that enable the robot to do drilling, visual servoing of a moving object, and writing his name on the white-board.

There are a wide variety of training approaches for spiking neural networks for neuromorphic deployment. However, it is often not clear how these training algorithms perform or compare when applied across multiple neuromorphic hardware platforms and multiple datasets. In this work, we present a software framework for comparing performance across four neuromorphic training algorithms across three neuromorphic simulators and four simple classification tasks. We introduce an approach for training a spiking neural network using a decision tree, and we compare this approach to training algorithms based on evolutionary algorithms, back-propagation, and reservoir computing. We present a hyperparameter optimization approach to tune the hyperparameters of the algorithm, and show that these optimized hyperparameters depend on the processor, algorithm, and classification task. Finally, we compare the performance of the optimized algorithms across multiple metrics, including accuracy, training time, and resulting network size, and we show that there is not one best training algorithm across all datasets and performance metrics.

Autonomous systems increasingly rely on on-board computation to avoid the latency overheads of offloading to more powerful remote computing. This requires the integration of hardware accelerators to handle the complex computations demanded by date-intensive sensors. FPGAs offer hardware acceleration with ample flexibility and interfacing capabilities when paired with general purpose processors, with the ability to reconfigure at runtime using partial reconfiguration. Managing dynamic hardware is complex and has been left to designers to address in an ad-hoc manner without first-class integration in autonomous software frameworks. This paper presents an abstracted runtime for managing adaptation of FPGA accelerators, including partial reconfiguration and parametric changes, that presents as a typical interface used in autonomous software systems. We present a demonstration using the Robot Operating System (ROS), showing negligible latency overhead as a result of the abstraction.

ROS has been completely refactored and evolved into ROS 2 to address the ever-increasing software complexity of autonomous machines. While it has become a de facto standard software platform for autonomous machines, ROS 2 still has room for improvement. It lacks support for essential features such as real-time stream processing, mode change, sensor fusion and rate control for output shaping. Moreover, its programming model is at a lower level than what most programmers would expect.In this paper, we carefully analyze the shortcomings of ROS 2 and propose to augment it with the Splash programming framework. In doing so, we host Splash on top of the ROS 2 software stack by conducting model conversion between Splash and ROS 2. We refer to the end result as Splash on ROS 2. To show its viability, we conducted a case study with a robot arm controller performing DNN-based object detection and motion planning. The case study qualitatively confirms that Splash on ROS 2 relieves the programming burden on developers, increases the software development productivity and improves the quality of the software.

This paper presents the Earth Observation Simulator (EO-Sim), a software framework which facilitates the design of novel observation systems. EO-Sim allows exploration of observing strategies by facilitating users to configure and simulate heterogenous satellite constellations. A set of potential observation opportunities and the associated observation metrics during mission-operations can be generated by the simulations. EO-Sim also incorporates an observation simulator to mock the operation of instruments taking into consideration the instrument specifications and observation geometry. A beta version has been made available to the public.

Human-robot collaboration and cooperation are critical for Autonomous Mobile Robots (AMRs) in order to use them in indoor environments, such as offices, hospitals, libraries, schools, factories, and warehouses. Since a long transition period might be required to fully automate such facilities, we have to deploy AMRs while improving safety in the mixed environments of human and mobile robots. In addition, human behaviors in such environments might be difficult to predict. In this paper, we present a Digital Twin for Autonomous Mobile Robots system named DigiMobot to support, manage, monitor, and validate AMRs in indoor environments. First, DigiMobot can simulate human behaviors and robot movements to verify and validate AMRs to improve safety in a virtual world. Secondly, DigiMobot can monitor and manage AMRs in the physical world by collecting sensor data from each robot in real-time. Since DigiMobot enables us to test the robot systems in the virtual world, we can deploy and implement AMRs in each facility without any modifications. To show the feasibility of DigiMobot, we develop a software framework and two different types of autonomous mobile robots. Finally, we conduct real-world experiments in a warehouse located in Saitama, Japan, in which more than 400, 000 items are stored for commercial purposes.

Coarse-grained reconfigurable arrays (CGRAs) are programmable hardware platforms that can be used to realize application-specific accelerators for higher performance and energy efficiency. A CGRA is a 2D array of configurable logic blocks & interconnect, where the logic blocks are typically large & ALU-like, and the interconnect is word-wide. CGRA-ME is a software framework that enables the modelling and exploration of CGRA architectures, as well as research on CGRA CAD algorithms. With CGRA-ME, an architect can specify a CGRA architecture at a high level of abstraction. A set of applications can be mapped onto the architecture to assess the mappability, power, performance and cost. CGRA-ME also allows one to generate synthesizable Verilog RTL for the modelled CGRA, permitting its implementation as an ASIC or FPGA overlay. In this paper, we describe the CGRA-ME framework [5] and overview its capabilities and current limitations. We discuss ongoing and prior research conducted with the framework, as well as outline future plans. We believe CGRA-ME will be a valuable contribution to the community, enabling new research on CGRA CAD & architectures.

Vibration monitoring uses data gathered from accelerometers to study kinetic phenomena in applications such as: structural health monitoring and predictive maintenance. The Internet of Things (IoT) has the potential to greatly expand the range and scope of vibration monitoring applications by delivering long-life wireless sensors that can be cost-effectively embedded in hard to reach places such as; within machines, infrastructure or the built environment. However, achieving this vision is difficult due to the stringent resource constraints of contemporary IoT devices and networks. This has led the research community to develop a creative range of application-specific near-sensor processing firmware. However, systematic support for generic vibration monitoring on resource-poor IoT networks remains an open problem. We tackle this challenge by introducing ReFrAEN, a software framework that efficiently enables a wide range of vibration monitoring applications on IoT networks. ReFrAEN achieves this through a deeply configurable combination of compression techniques and data processing algorithms. These features allow end-users to effectively trade-off between resource consumption and data resolution in order to meet battery life constraints while preserving sufficient data quality to support the target application. Our evaluation shows that ReFrAEN is capable of identifying bearing faults, while dramatically improving battery lifetime and reducing latency in comparison to prior approaches.

This paper presents a methodology for strategic day-ahead planning that uses a combination of deep learning and optimization. A noise-driven recurrent neural network structure is proposed for forecasting electricity prices and local inflow to water reservoirs. The resulting forecasters generate predictions with seasonal variation without relying on long input sequences. This forecasting method is employed in a stochastic program formulation of the day-ahead problem. This results in optimal order strategies for a price-taking hydropower producer participating in the Nordic day-ahead market. Using an open-source software framework for stochastic programming, the model is implemented and distributed over multiple cores. The model is then solved in parallel using a sampling-based algorithm. Tight confidence intervals around the stochastic solution are provided, which show that the gain from adopting a stochastic approach is statistically significant.

Computer-aided design (CAD) software is a fundamental engineering tool that is often integrated into educational programs, including those for electronics. When the COVID-19 pandemic spread worldwide in 2020, an introductory electronics course at our Institution had to be reformulated due to restrictions on holding face-to-face classes. In this scenario, traditional lab tasks were replaced by computer-based assignments that students worked on using CAD software. This paper introduces the software framework used for this purpose, which has the two features considered necessary to closely replicate a practical experiment: i) the ability to mimic breadboard circuit prototyping activities and ii) the ability to simulate the prototyped circuits. In order to enhance students' motivation for the educational activities, the assignments proposed in our course have focused on robotics applications for several years. This was conveniently accounted for by combining simulation of electronic circuits with robotics simulation. In this way, we can be reproduce the embodiment of electronic circuits and the interaction with an external environment that takes place in a real experiment. The paper describes the proposed software framework, gives examples of its use, and reports on students' results for the first distance learning semester using the framework.

With the recent advances and the proliferation of low-power and resource constraint embedded hardware being deployed in the field, the edge computing paradigm has become more prominent. Nowadays, it is possible to apply Machine Learning (ML) algorithms and Deep learning (DL) architectures, like Deep Neural Networks (DNNs), by exploiting these edge devices. There is also a plethora of software frameworks to transform the typically complex computations and accommodate the implementation process and specifically for those who are targeted to field-programmable gate arrays (FPGAs), quite a few challenges need to be addressed. Therefore, there is an active research development towards providing efficient techniques for applying DNNs models to these devices. In this paper, we propose a methodology which combines two research tools for the generation of a register-transfer Level (RTL) design using a hardware description language (HDL). We evaluate different implementations experimenting with a set of DNN models in different FPGA boards.

Galaxy is an open-source web-based framework that is widely used for performing computational analyses in diverse application domains, such as genome assembly, computational chemistry, ecology, and epigenetics, to name a few. The current Galaxy software framework runs on several high-performance computing platforms such as on-premise clusters, public data centers, and national lab supercomputers. These infrastructures also provide support for state-of-the-art accelerators like Graphical Processing Units (GPUs). When coupled with accelerator support, the tools executing in Galaxy can benefit from massive performance gains in terms of computation time, thereby allowing a more robust computational analysis environment for researchers. Despite tools having GPU capabilities, the current Galaxy framework does not support GPUs, and thus prevents tools from taking advantage of the performance benefits offered by GPUs. We present and experimentally evaluate GYAN, a GPU-aware computation mapping and orchestration functionality implemented in Galaxy that allows the Galaxy tools to be executed on a GPU-enabled cluster. GYAN has the capability of identifying GPU-supported tools and scheduling them on single or multiple GPU nodes based on the availability in the cluster. GYAN supports both native and containerized tool execution. We performed extensive evaluations of the implementation using popular bio-engineering tools to demonstrate the benefits of using GPU technologies. For example, the Racon consensus tool executes ~2× faster than the regular baseline CPU-only jobs, while the Bonito base calling tool shows ~50× speedup.

Computational chemistry is the leading application to demonstrate the advantage of quantum computing in the near term. However, large-scale simulation of chemical systems on quantum computers is currently hindered due to a mismatch between the computational resource needs of the program and those available in today's technology. In this paper we argue that significant new optimizations can be discovered by co-designing the application, compiler, and hardware. We show that multiple optimization objectives can be coordinated through the key abstraction layer of Pauli strings, which are the basic building blocks of computational chemistry programs. In particular, we leverage Pauli strings to identify critical program components that can be used to compress program size with minimal loss of accuracy. We also leverage the structure of Pauli string simulation circuits to tailor a novel hardware architecture and compiler, leading to significant execution overhead reduction by up to 99%. While exploiting the high-level domain knowledge reveals significant optimization opportunities, our hardware/software framework is not tied to a particular program instance and can accommodate the full family of computational chemistry problems with such structure. We believe the co-design lessons of this study can be extended to other domains and hardware technologies to hasten the onset of quantum advantage., Comment: 12 pages, 11 figures, to appear in ISCA 2021

Coronavirus disease 2019 (COVID-19) has resulted in an ongoing pandemic. Since COVID-19 spreads mainly via close contact among people, social distancing has become an effective manner to slow down the spread. However, completely forbidding close contact can also lead to unacceptable damage to the society. Thus, a system that can effectively monitor people’s social distance and generate corresponding alerts when a high infection probability is detected is in urgent need.In this paper, we propose SmartDistance, a smartphone based software framework that monitors people’s interaction in an effective manner, and generates a reminder whenever the infection probability is high. Specifically, SmartDistance dynamically senses both the relative distance and orientation during social interaction with a well-designed relative positioning system. In addition, it recognizes different events (e.g., speaking, coughing) and determines the infection space through a droplet transmission model. With event recognition and relative positioning, SmartDistance effectively detects risky social interaction, generates an alert immediately, and records the relevant data for close contact reporting. We prototype SmartDistance on different Android smartphones, and the evaluation shows it reduces the false positive rate from 33% to 1% and the false negative rate from 5% to 3% in infection risk detection.

In view of the problem that the existing urban and rural planning system does not establish enough three-dimensional concept level in the process of planning database, which leads to excessive system noise, the urban and rural planning system is designed based on remote sensing and geographic information system. In the hardware design of urban and rural planning system, planning system circuit and A/D, D/A conversion chip, in the software design of urban and rural planning system, planning software framework, the establishment of system database hierarchy model, improve the function structure, realize the optimization of urban and rural planning system. In the experiment, the noise processing ability of the system is compared with the existing three systems. According to the experimental data, the average coefficient of the system in ten experiments is 0.2958, and the average coefficients of the other three systems are 0.2305, 0.2213 and 0.2245. It can be seen that the system has better noise processing ability, and the system noise is small, which realizes the optimization design of urban and rural planning system.

In this paper, we present a design and first proof-of-concept test results of a distributed radio-localization system for tracking of moving targets with a swarm of drones. In particular, we develop and test a software framework that uses bistatic ranges to follow a target in-flight. We conduct full-system flight test with up to two drones in the air, three simulated drones, and a laptop equipped with a GPS module as the target. The system is shown to dynamically track the target as it moved, with both real and simulated drones following it in a specified formation.

Many popular deep neural networks implement an input-output mapping that is non-smooth with respect to the network parameters. This non-smoothness may have contributed to the difficulty of analyzing deep learning theoretically. Sophisticated approaches have recently been proposed to address this specific difficulty. In this note, we explore a simple approach consisting instead in smoothing the input-output mapping. We show how to perform smoothing automatically within a differentiable programming framework. The impact of the smoothing on the convergence behavior can then be automatically controlled. We illustrate our approach with numerical examples using multilayer perceptrons.

Self-sufficient, robotic spacecraft require estimates of their hardware health state in order to project future system state and plan actions toward achieving mission goals. In this paper, we report on integration of a Model-Based Fault Diagnosis (MBFD) model and reasoning engine into flight software leveraging the Arcsecond Space Telescope Enabling Research in Astrophysics (ASTERIA) mission, including test results against captured flight data using the ASTERIA system testbed. Our effort integrated the Model-based Off-Nominal State Identification and Detection (MONSID) model-based reasoning system, developed by Okean Solutions, into ASTERIA flight software using the F Prime software framework. The MONSID engine was supplied with a model of the Blue Canyon Technologies XACT attitude control system (ACS) and tested against flight data and seeded fault tests. While we were unable to conduct an on-board experiment due to the premature loss of ASTERIA, our effort proved the feasibility of on-board model-based fault management, demonstrating reliable and accurate diagnosis using captured data, and further supporting a closed-loop spacecraft autonomy demonstration including autonomous navigation in off-nominal conditions.

Space missions have required the development of specific embedded systems for spacecraft applications, making reuse of existing software difficult. JAXA's space robot operating system (SP-ROS) initiative tackles this problem in order to realize more intelligent robotic space missions. JAXA's SP-ROS initiative proposes an open-source software framework for robots used in outer space and will facilitate the development of space robotics through the ROS and cFS system (RACS). This paper introduces RACS and describes a planned demonstration in space. RACS provides two functionalities. The first one is ROS/cFE converter, which makes it possible to directly incorporate the code of a ground robot application into an environment that can be ported as a spacecraft application. The porting of existing robotic applications will aid in developing new applications using ROS's rich development environment. The second functionality of RACS is the ROS/cFE bridge, which allows native cFS and ROS programs to interact. This is aimed mainly at making ground robotics software testing easier. Another benefit would allow ROS's development tools (Rviz and Gazebo) and libraries (OpenCV and MoveIt), as well as packages developed by the roboticists anywhere in the world, to be directly incorporated in the testing. Provided the onboard computer supports cFS, it will be possible to implement these software packages into space robots. JAXA has made the open-source RACS module available to everybody. We currently propose to test the system in a space demonstration using Int-Ball2, a space drone in the International Space Station (ISS).

Hadoop MapReduce is a software framework for processing vast amounts of data in parallel on large clusters. As data size increases, a need arises to resolve the significant increase in the disk $I/O$ of reducer nodes, which may cause runtime bottlenecks. In response, we propose Bucket MapReduce, a system that utilizes bucket sort and pipeline parallelism to improve the performance of Hadoop MapReduce. Several parameters may influence the performance in Bucket MapReduce, and thus we will further discuss parameter tuning. In this paper, we perform experiments on TeraSort benchmark. Bucket MapReduce successfully reduces local disk $I/O$ by 61% and improves the runtime by 1. 39x in 800GB TeraSort benchmark.

Software frameworks make developing applications for a specific domain easier than doing so from scratch. Unfortunately, frameworks can also place unexpected requirements on a developer’s application, which can, in turn, lead to application bugs in the development process. We propose an automated technique for repairing violations of state-based framework requirements, FrameFix. First, developers of a framework can encode state-based framework requirements. Then, FrameFix automatically checks whether a developer’s application follows the encoded requirements. Once a violation of these requirements has been detected, FrameFix tries three different approaches — reordering method calls, moving method calls to different method definitions, and comparing the faulty method to similarly defined methods on GitHub. These repair approaches are based on the principles of how frameworks interact with framework applications: object protocols and inversion of control. To demonstrate that these principles can be used to aid automated repair for framework applications, we created a sample implementation of FrameFix for Android applications. Our evaluation shows that FrameFix is effective, repairing both real bugs in real applications, and a large number and variety of injected defects.