Autonomous Applications Research Articles

Architectural pattern for service collaboration

The aim of this paper is to propose a modeling framework, tailored to build efficient, elastic and autonomous applications from tasks and services. It includes integrated services to develop the software products, reusing on demand in-house services with specific requirements and flexible the representational state transfer (REST) services. The idea is to decouple authorization for reduced service dependency and to provide a possibility for developing the whole application by increasing the existing application flexibility. Based on the fact that there are different web application platforms that serve to offer services to users but they are not integrated; we propose a framework with high flexibility degree, especially integrating the most used services such: e-learning, administrative, and library services, as University services are concern.

Trajectory Tracking via Backstepping Controller with PID or SMC for Mobile Robots

Mobile robot concept is one of the most commonly used nonholonomic system for industrial and academic autonomous applications. There are many types of mobile robot design concepts and control strategies which have been continuously developed by researchers. In this study, two wheeled differential drive mobile robot (DDMR) is used for trajectory tracking study under different conditions. Reference trajectory, dynamic and kinematic motion models of DDMR are defined as mathematical expressions in computer software. For tracking the reference trajectory, error between current pose and reference pose was decreased by sliding mode controller (SMC) and proportional–integral–derivative controller (PID) with kinematic based backstepping controller (KBBC) respectively. A reference path which consists of sinusoidal and linear parts tracked by both controller combinations in first simulation to examine controller tracking performances. In order to examine and compare; responsiveness, stability and robustness of the controllers, an additional mass which affects motion dynamics of DDMR vertically added to the mobile robot body during trajectory tracking application. All results and discussions are comparatively stated at the end of the study with related error figures and evaluations.

Feasibility of Using Poisons to Suppress the Positive Temperature Reactivity Coefficient in Hydride-Moderated Reactors

Metal hydrides are being seriously considered for advanced nuclear reactor or microreactor applications due to their solid physical state and high hydrogen density. Using hydrides for autonomous applications poses several research and development challenges, one of which relates to neutron upscattering in the thermal energy regime. These hydrides, including zirconium hydride and yttrium hydride, result in a positive temperature coefficient of reactivity for several advanced reactor designs. In this study, we consider one such design that exhibits positive feedback from metal hydrides and thoroughly investigate the neutronic aspects of the core. Temperature reactivity coefficients for four fuels and two hydride moderator configurations are studied, and the total temperature coefficients are found to be positive for all designs, showing that this issue cannot be resolved simply by material variations. Accordingly, five epi-thermal absorbers were evaluated to demonstrate the feasibility of the excess positive feedback suppression in the core instigating from neutron energy spectrum shift. Following which, two promising burnable poison candidates are selected to investigate further throughout the core discharge. Promising results are shown for this core design, which can be extended to other hydride-moderated remote special-purpose reactor designs.

Low‐power temperature‐independent CMOS measurement circuits for resistive gas sensors

SummaryIn this work, we present the design of three different circuits for the measurement of resistive gas sensors. Schematic designs of each approach are discussed and simulated using HSPICE in a commercial CMOS 180 nm technology node. Their physical layout is also implemented. The three designs output a pulsed voltage signal whose period is modulated by the resistance of the sensor. They also implement differential measurement techniques in order to reduce the effect of temperature variations. The first design uses an operational amplifier and a resistor to measure the sensor and transduce its signal; the second one employs a Wheatstone bridge with common‐mode feedback and a bulk‐input OTA; and the third one utilizes a current‐mode Wheatstone bridge built with a current‐conveyor 2. The designs consume between 1.8 and W; their temperature coefficient is always between 60 and 305 ppm/° C; and their resolution is between 0.25% and 1%. The circuits are compatible with sensors of a wide resistance range, being the best one tolerant to resistances between 2 M and 1 G . Therefore, these designs are ideal for low‐power autonomous applications in environments where temperature is not controlled.

OCDMA Codeword Switching Technique to Avoid Interference of Time-of-Flight LiDAR System for Autonomous Vehicles

The direct time-of-flight (d-ToF) technique is the most promising method for light detection and ranging (LiDAR) systems in long-range autonomous applications. However, the receiver cannot distinguish its own pulse signal from those of other LiDAR systems. This causes mutual interference between LiDAR systems, which degrades object recognition. The optical code division multiple access (OCDMA) modulation technique can be used to avoid mutual interference. However, it is difficult to secure sufficient orthogonal codes, because each laser channel uses a designated codeword. In addition, when interference occurs in a specific direction, it can occur continuously. Adversarial attacks on a LiDAR system can trick a vehicle to perform false object recognition without physical contact. This can endanger passengers, pedestrians, and other vehicles. Therefore, we propose an OCDMA codeword switching technique that can effectively solve these problems in driving environments. First, we analyzed the characteristics of optical orthogonal codes (OOC) and defined the interference generated in LiDAR systems. The proposed technique exhibits interference robustness by using a code index buffer based on differences in the received optical signal power. In addition, verification of the hardware implementation and its complexity was analyzed using LiDAR testbench and an FPGA. The interference and continuous interference of the proposed technique in various driving environments were verified using simulation model. Compared with using only a designated codeword, the proposed technique reduces interference by 82.59% and 85.51%, and continuous interference by 92.24% and 96.94% in straight-line and curved driving, respectively.

A blockchain-based platform for trading weather derivatives

This study investigated the technical and economic viability of hedging electricity consumption using weather derivatives and smart contracts. For this purpose, we priced call options to hedge against excess of temperature for 5 Brazilian cities. We also developed a distributed autonomous application (DApp) using smart contracts, which allows individuals to negotiate these financial instruments. Our findings suggest that blockchain technology can be useful in providing a low-cost infrastructure to develop financial instruments to hedge weather-related losses. The cost of providing this platform has been estimated to be less than 300 USD. The price of options has been estimated under 50 USD. This is particularly useful for electricity consumers and small businesses in poor countries. In addition, this study provides a comprehensive guide for the development of financial solutions using smart contracts to mitigate climate change impacts.

OpTIFlow – An optimized end-to-end dataflow for accelerating deep learning workloads on heterogeneous SoCs

A typical edge compute SoC capable of handling deep learning workloads at low power is usually heterogeneous by design. It typically comprises multiple initiators such as real-time IPs for capture and display, hardware accelerators for ISP, computer vision, deep learning engines, codecs, DSP or ARM cores for general compute, GPU for 2D/3D visualization. Every participating initiator transacts with common resources such as L3/L4/DDR memory systems to seamlessly exchange data between them. A careful orchestration of this dataflow is important to keep every producer/consumer at full utilization without causing any drop in real-time performance which is critical for automotive applications. The software stack for such complex workflows can be quite intimidating for customers to bring-up and more often act as an entry barrier for many to even evaluate the device for performance. In this paper we propose techniques developed on TI’s latest TDA4V-Mid SoC, targeted for ADAS and autonomous applications, which is designed around ease-of-use but ensuring device entitlement class of performance using open standards such as DL runtimes, OpenVx and GStreamer.

Identifying Novel Defects during AI-driven Visual Quality Inspection

Visual Quality Inspection is an integral part of the manufacturing process, becoming increasingly automated with the advent of Industry 4.0. While modern industries reap substantial benefits from using AI-driven Computer Vision Algorithms and Deep Neural Networks, these technologies bring about new issues, such as the requirement for extensive training datasets and the lack of robustness to out-of-distribution samples. This can be particularly risky in real-world autonomous applications such as cyber-physical reinforcement learning systems where a lack of resilience to unexpected inputs could compromise the integrity of the production process or put human operators at risk. Our research explicitly examines scenarios of handling unanticipated defect types during the continuous operation of Visual Quality Inspection systems. To this end, many modern Machine Learning methods can be employed, such as open-set recognition, semi-supervised novelty detection, and intelligent data augmentation. We examine a range of proven and state-of-the-art methods focusing on Deep Learning over a real-world dataset. Our comparative analysis highlights significant trade-offs between approaches and provides concrete suggestions for immediate implementation and further research toward robust Visual Quality Inspection.

A Review of UAV Platforms for Autonomous Applications: Comprehensive Analysis and Future Directions

A Review of UAV Platforms for Autonomous Applications: Comprehensive Analysis and Future Directions

Stereopsis without correspondence.

Stereopsis has traditionally been considered a complex visual ability, restricted to large-brained animals. The discovery in the 1980s that insects, too, have stereopsis, therefore, challenged theories of stereopsis. How can such simple brains see in three dimensions? A likely answer is that insect stereopsis has evolved to produce simple behaviour, such as orienting towards the closer of two objects or triggering a strike when prey comes within range. Scientific thinking about stereopsis has been unduly anthropomorphic, for example assuming that stereopsis must require binocular fusion or a solution of the stereo correspondence problem. In fact, useful behaviour can be produced with very basic stereoscopic algorithms which make no attempt to achieve fusion or correspondence, or to produce even a coarse map of depth across the visual field. This may explain why some aspects of insect stereopsis seem poorly designed from an engineering point of view: for example, paying no attention to whether interocular contrast or velocities match. Such algorithms demonstrably work well enough in practice for their species, and may prove useful in particular autonomous applications. This article is part of a discussion meeting issue 'New approaches to 3D vision'.

Muller C-Element Exploiting Programmable Metallization Cell for Bayesian Inference

Decision-making via Bayesian inference is a prominent operation in several autonomous applications, including robotics, brain-machine interactions, artificial intelligence (AI) agents, etc. A cascaded tree of asynchronous logic elements, known as Muller C-elements, can perform stochastic Bayesian inference efficiently. Therefore, there is an urgent need for compact and energy-efficient Muller C-element realizations. To this end, in this work, for the first time, we propose a Muller C-element utilizing a Programmable Metallization Cell (PMC) and a CMOS inverter. Using an experimentally calibrated, in-house developed physics-based compact model of Ag-Ge0.3Se 0.7 PMC and CMOS inverter (in 7 nm technology node), we show that the proposed Muller C-element is extremely compact and consumes at least <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\sim 9\times $ </tex-math></inline-formula> less power compared to the previously reported implementations. Furthermore, we demonstrate that the proposed PMC-based Muller C-element can make Bayesian inferences on bitstream-encoded data with reasonable accuracy. Our results indicate that the computational precision depends significantly on the number of output bits and the probability encoded by the inputs to the Muller C-element. Moreover, we also propose a novel readout circuit design to facilitate cascading of multiple PMC-based Muller C-elements to increase the number of evidence sources and demonstrate its efficacy for spam filtering applications.

Autonomous urban topology generation for urban flow modelling

• New framework is developed that integrates the autonomous urban topology generation for urban flow modeling. • A new deep learning model is developed for building footprints extraction from satellite imagery • Deep learning model was meshed with Light Detection and Ranging (LiDAR) data by meshing to generate 3D building models • An illustration pedestrian level wind simulation was carried out for city of London On. One of the challenges in realistic numerical urban micro-climate modelling for wind, heat transfer, and building energy simulation applications is the complexity of urban topology and complex building geometries. This paper presents a deep learning modelling for building footprint polygon extraction from satellite imagery that is integrated with Light Detection and Ranging (LiDAR) data to generate 3D building models. The deep learning model registered a mean squared error of 0.02. The trained deep learning model can then be applied to a new set of input image data to extract building footprint polygons for autonomous application, and it can also be incrementally retrained with good quality data when it becomes available. A framework is developed that integrates the autonomous urban topology generator with urban flow modelling. The modelling steps are explained through an application example of urban flow modelling encountered during a pedestrian level wind assessment for the city of London, Ontario.

Automating Global Threat-Maps Generation via Advancements of News Sensors and AI

Negative events are prevalent all over the globe round the clock. People demonstrate psychological affinity to negative events, and they incline to stay away from troubled locations. This paper proposes an automated geospatial imagery application that would allow a user to remotely extract knowledge of troubled locations. The autonomous application uses thousands of connected news sensors to obtain real-time news pertaining to all global troubles. From the captured news, the proposed application uses artificial intelligence-based services and algorithms like sentiment analysis, entity detection, geolocation decoder, news fidelity analysis, and decomposition tree analysis to reconstruct global threat maps representing troubled locations interactively. The fully deployed system was evaluated for full three months of summer 2021, during which the autonomous system processed above 22 k news from 2397 connected news sources involving BBC, CNN, NY Times, Government websites of 192 countries, and all possible major social media sites. The study revealed 11,668 troubled locations classified successfully with outstanding precision, recall, and F1-score, all evaluated in ubiquitous environment covering mobile, tablet, desktop, and cloud platforms. The system generated interesting global threat maps for robust scenario set of 3.71 times {10}^{29}, to be reported as original fully autonomous remote sensing application of this kind. The research discloses attractive news and global threat-maps with trusted overall classification accuracy.

CNS Flight Stack for Reproducible, Customizable, and Fully Autonomous Applications

While low-level auto pilot stacks for aerial vehicles focus on robust control, sensing, and estimation, the continuous advancement of higher-level autonomy for aerial vehicles requires much more complex higher-level flight stacks in order to enable safe, fully autonomous long-duration missions. Rather than focusing on the low-level control, high-level flight stacks are required to monitor the system's integrity continuously, initiate contingency plans, execute mission plans and adapt them in non-nominal situations, allow for proper data logging, and provide standardized interfaces and integrity verification for external mission planners and localization modules. To that end, we present our freely available, high-level flight stack dubbed CNS Flight Stack that meets the above requirements and at the same time a) is platform-agnostic through a generalized (embedded) hardware abstraction layer, b) uses low compute complexity for online use on embedded hardware, and c) can be extended with other sensor modalities, integrity checks, and mission modules. These additional properties make it reproducible on a variety of different platforms for safe and fully autonomous applications. We tested the proposed flight stack in over 450 real-world flights and report the failure modes our framework detected and also mitigated to avoid crashes of the aerial system.

Magnets-assisted dual-mode triboelectric sensors integrated with an electromagnetic generator for self-sustainable wireless motion monitoring systems

Self-powered motion sensors are demanded as an energy-efficient technology for future autonomous applications. However, realizing fully self-powered random motion-sensing at a system level is still challenging. Herein, the dual-mode triboelectric nanogenerators (TENGs) are integrated with an electromagnetic generator (EMG) to realize a self-sustained wireless random motion-sensing system. For EMG, a magnetic repulsion mechanism is introduced which equally assists the self-actuation of sliding TENG (S-TENG) and contact separation TENG (CS-TENG) simultaneously. To accomplish self-actuation on CS-TENG, a double-layered flexible thin film of PDMS-FeSiCr/PDMS is used which can simultaneously work as an active TENG layer as well as actuating layer thereby reducing the operational complexity. The motion-sensing resolution and sensitivity of S-TENG are improved by implementing interdigitated electrodes and surface modifications, whereas the self-actuating CS-TENGs offer precise detection of motion directions. The dual-mode TENGs combinedly can measure random motion parameters and motion directions along linear, rotational, and tilting conditions accurately. Moreover, the self-powered motion sensors exhibit excellent acceleration (1.966 V.s2/m), frequency (14.099 V/Hz), and tilting angle (0.257 V/degree) sensitivity with greater accuracy. The electricity generated from EMG (peak power: 103.88 mW) is sufficient to operate signal processing and transmission components for self-powered wireless control of the robotic ball balancing platform, showing future possibilities for self-powered autonomous control applications.

Longitudinal Predictive Control for Vehicle-Following Collision Avoidance in Autonomous Driving Considering Distance and Acceleration Compensation.

In response to the widespread adoption of vehicle-following systems in autonomous applications, the demand for collision warning to enable safer functionalities is increasing. This study provides an approach for automated vehicle guidance to follow the preceding vehicles longitudinally and puts emphasis on the performance of collision avoidance. The safety distance model is established, which contains a distance compensation algorithm to deal with the special case on curved roads. By introducing the algorithm of velocity and distance prediction, the collision risks are detected and measured in real time. The objective function is established based on optimal control theory to solve the desired following acceleration. The control system designed with the method of proportion integration differentiation combines throttle percentage and brake pressure as outputs to compensate acceleration. In the Carsim and Simulink co-simulation platform, the control system for longitudinal collision avoidance is simulated and analysed for four typical working conditions: the preceding vehicle drives at a constant speed on straight and curved roads, while the preceding vehicle drives at various speeds on straight and curved roads. The results validate the feasibility and effectiveness of the proposed method, which can be used for the longitudinal control of vehicle-following active collision avoidance.

Non-Invasive Data Acquisition and IoT Solution for Human Vital Signs Monitoring: Applications, Limitations and Future Prospects.

The rapid development of technology has brought about a revolution in healthcare stimulating a wide range of smart and autonomous applications in homes, clinics, surgeries and hospitals. Smart healthcare opens the opportunity for a qualitative advance in the relations between healthcare providers and end-users for the provision of healthcare such as enabling doctors to diagnose remotely while optimizing the accuracy of the diagnosis and maximizing the benefits of treatment by enabling close patient monitoring. This paper presents a comprehensive review of non-invasive vital data acquisition and the Internet of Things in healthcare informatics and thus reports the challenges in healthcare informatics and suggests future work that would lead to solutions to address the open challenges in IoT and non-invasive vital data acquisition. In particular, the conducted review has revealed that there has been a daunting challenge in the development of multi-frequency vital IoT systems, and addressing this issue will help enable the vital IoT node to be reachable by the broker in multiple area ranges. Furthermore, the utilization of multi-camera systems has proven its high potential to increase the accuracy of vital data acquisition, but the implementation of such systems has not been fully developed with unfilled gaps to be bridged. Moreover, the application of deep learning to the real-time analysis of vital data on the node/edge side will enable optimal, instant offline decision making. Finally, the synergistic integration of reliable power management and energy harvesting systems into non-invasive data acquisition has been omitted so far, and the successful implementation of such systems will lead to a smart, robust, sustainable and self-powered healthcare system.

How to Implement Automotive Fault Diagnosis Using Artificial Intelligence Scheme.

The necessity of vehicle fault detection and diagnosis (VFDD) is one of the main goals and demands of the Internet of Vehicles (IoV) in autonomous applications. This paper integrates various machine learning algorithms, which are applied to the failure prediction and warning of various types of vehicles, such as the vehicle transmission system, abnormal engine operation, and tire condition prediction. This paper first discusses the three main AI algorithms, such as supervised learning, unsupervised learning, and reinforcement learning, and compares the advantages and disadvantages of each algorithm in the application of system prediction. In the second part, we summarize which artificial intelligence algorithm architectures are suitable for each system failure condition. According to the fault status of different vehicles, it is necessary to carry out the evaluation of the digital filtering process. At the same time, it is necessary to preconstruct its model analysis and adjust the parameter attributes, types, and number of samples of various vehicle prediction models according to the analysis results, followed by optimization to obtain various vehicle models. Finally, through a cross-comparison and sorting, the artificial intelligence failure prediction models can be obtained, which can correspond to the failure status of a certain car model and a certain system, thereby realizing a most appropriate AI model for a specific application.

An Optimized Path-Planning Concept for a Micro-Mouse Robot for Autonomous Applications in Terrained Environments

An Optimized Path-Planning Concept for a Micro-Mouse Robot for Autonomous Applications in Terrained Environments

MiniNet: Dense squeeze with depthwise separable convolutions for image classification in resource-constrained autonomous systems

In recent years, artificial intelligence (AI) has been developed vigorously, and a great number of AI autonomous applications have been proposed. However, how to decrease computations and shorten training time with high accuracy under the limited hardware resource is a vital issue. In this paper, on the basis of MobileNet architecture, the dense squeeze with depthwise separable convolutions model is proposed, viz. MiniNet. MiniNet utilizes depthwise and pointwise convolutions, and is composed of the dense connection technique and the Squeeze-and-Excitation operations. The proposed MiniNet model is implemented and experimented with Keras. In experimental results, MiniNet is compared with three existing models, i.e., DenseNet, MobileNet, and SE-Inception-Resnet-v1. To validate that the proposed MiniNet model is provided with less computation and shorter training time, two types as well as large and small datasets are used. The experimental results showed that the proposed MiniNet model significantly reduces the number of parameters and shortens training time efficiently. MiniNet is superior to other models in terms of the lowest parameters, shortest training time, and highest accuracy when the dataset is small, especially.