Search

Computationally intensive Inference tasks of Deep neural networks have enforced revolution of new accelerator architecture to reduce power consumption as well as latency. The key figure of merit in hardware inference accelerators is the number of multiply-and-accumulation operations per watt (MACs/W), where, the state-of-the-arts MACs/W remains several hundreds Giga-MACs/W. We propose a Tera-MACS/W neural hardware inference Accelerator (TMA) with 8-bit activations and scalable integer weights less than 1-byte. The architectures main feature is configurable neural processing element for matrix-vector operations. The proposed neural processing element has Multiplier-less Massive Parallel Processor to work without any multiplications, which makes it attractive for energy efficient high-performance neural network applications. We benchmark our systems latency, power, and performance using Alexnet trained on ImageNet. Finally, we compared our accelerators throughput and power consumption to the prior works. The proposed accelerator outperforms the state of the art in terms of energy and area achieving 2.3 TMACS/W@1.0 V, 65 nm CMOS technology.

Faster adaptability to open-world novelties by intelligent agents is a necessary factor in achieving the goal of creating Artificial General Intelligence (AGI). Current RL framework does not considers the unseen changes (novelties) in the environment. Therefore, in this paper, we have proposed OODA-RL, a Reinforcement Learning based framework that can be used to develop robust RL algorithms capable of handling both the known environments as well as adaptation to the unseen environments. OODA-RL expands the definition of internal composition of the agent as compared to the abstract definition in the classical RL framework, allowing the RL researchers to incorporate novelty adaptation techniques as an add-on feature to the existing SoTA as well as yet-to-be-developed RL algorithms.

Fully autonomous vehicles (FAVs) lack monitoring inside the cabin. Therefore, an in-cabin monitoring system (IMS) is required for surveilling people causing irregular or abnormal situations. However, monitoring in the public domain allows disclosure of an individual’s face, which goes against privacy preservation. Furthermore, there is a contrary demand for privacy in the IMS of AVs. Therefore, an intelligent IMS must simultaneously satisfy the contrary requirements of personal privacy protection and person identification during abnormal situations. In this study, we proposed a privacy-preserved IMS, which can reidentify anonymized virtual individual faces in an abnormal situation. This IMS includes a step for extracting facial features, which is accomplished by the edge device (onboard unit) of the AV. This device anonymizes an individual’s facial identity before transmitting the video frames to a data server. We created different abnormal scenarios in the vehicle cabin. Further, we reidentified the involved person by using the anonymized virtual face and the reserved feature vectors extracted from the suspected individual. Overall, the proposed approach preserves personal privacy while maintaining security in surveillance systems, such as for in-cabin monitoring of FAVs.

The purpose of this research is to design a simple and intuitive path tracker to minimise the distance error of a car-like delivery service robot using only a position sensor, i.e. global positioning system in outdoor environments. The authors proposed not only a single-point preview path tracker (SPP), but also a curvature-based preview control using a quadratic curve fitting technique. This method can provide flexibility in preview distance control even at a constant speed, and reduces the burden of an actuator for maintenance. Furthermore, the performance of the proposed adaptive SPP is considerably improved compared to that when using a fixed preview distance.

We propose an artificial deep neural network- (ANN-) based automatic parking controller that overcomes a stubborn restriction prevalent in traditional approaches. The proposed ANN learns human-like control laws for automatic parking through supervised learning from a training database generated by computer-aided optimizations or real experiments. By learning the relationships between the instantaneous vehicle states and the corresponding maneuver parameters, the proposed twin controller yields lateral and longitudinal maneuvering parameters for executing automatic parking tasks in confined spaces. The proposed automatic parking controller exhibits a twin architecture comprising a main agent and its cloned agent. Before the main agent assumes a maneuvering action, the cloned agent predicts the consequences of the maneuvering action through a Collision Checking and Adjustment (CCA) system. The proposed parking agent operates like a human driver in a manner that is characterized by an unplanned trajectory. In addition, the kinematics of the subject vehicle is not exactly modelled for parking control. The simulation results demonstrate that the proposed twin agent emulates the attributes of a human driver such as adaptive control and determines the consequences of the tentative maneuvering action under varying kinematic models of the subject vehicle. We validate the proposed parking controller by simulating the software-in-the-loop architecture using a PreScan simulator in which the dynamics of the virtual vehicle’s behavior resemble a real vehicle.

We propose a noninvasive method for determining Langer’s tension lines and tracking two-dimensional (2D) deformation of human skin utilizing spatially resolved diffuse reflectance (SRDR) of infrared (IR) light measurement in vivo. The anisotropic light propagation pattern of the skin was measured at several locations on the human body; then, the principal axis (i.e., eigenvector) of the radial plot of equi-distant intensity lines was obtained to determine the orientation of the scattering-induced anisotropic reflectance of IR light. The results indicate the proposed method is able to determine preferential orientation of filamentous fiber with a reasonable degree of prediction accuracy over the entire body. We validated the proposed method by comparing the detected orientations of fiber with a previous scattering model-based light propagation anisotropy analysis method. We measured temperature dependence and tensile stress-induced changes of the spatially resolved reflectance of IR light on the skin. The surface temperature affects the light transmittance almost isotopically on the propagation patterns. The main contribution of this work is tracking the direction of tension causing 2D deformation of human skin without visible markers. The proposed method is applicable to sensing facial gestures or skin movements induced by human motion of skeletal or facial muscles.

The recent development and adoption of unmanned aerial vehicles (UAVs) is due to its wide variety of applications in public and private sector from parcel delivery to wildlife conservation. The integration of UAVs, 5G, and satellite technologies has prompted telecommunication networks to evolve to provide higher-quality and more stable service to remote areas. However, security concerns with UAVs are growing as UAV nodes are becoming attractive targets for cyberattacks due to enormously growing volumes and poor and weak inbuilt security. In this paper, we propose a UAV- and satellite-based 5G-network security model that can harness machine learning to effectively detect of vulnerabilities and cyberattacks. The solution is divided into two main parts: the model creation for intrusion detection using various machine learning (ML) algorithms and the implementation of ML-based model into terrestrial or satellite gateways. The system identifies various attack types using realistic CSE-CIC IDS-2018 network datasets published by Canadian Establishment for Cybersecurity (CIC). It consists of seven different types of new and contemporary attack types. This paper demonstrates that ML algorithms can be used to classify benign or malicious packets in UAV networks to enhance security. Finally, the tested ML algorithms are compared for effectiveness in terms of accuracy rate, precision, recall, F1-score, and false-negative rate. The decision tree algorithm performed well by obtaining a maximum accuracy rate of 99.99% and a minimum false negative rate of 0% in detecting various attacks as compared to all other types of ML classifiers.

The recent growth and adoption of Unmanned Aerial Vehicles (UAVs) is due to its low development cost, high aerial mobility, advanced battery technology, rotors, gyroscopes, GPS, cameras, sensors and a wide range of applications. The UAV offers new possibilities for business in civil and non-civil applications such as parcel delivery, aerial mapping, agriculture, wildlife conservation, surveillance, search and rescue, etc. However, considering the exponential growth in various types of aerial vehicles, it is clear that advanced air traffic management services are urgently needed to control and handle the ever increasing air traffics. In this paper, we first present the two major low altitude UAV traffic management system in the USA and Europe and compare them thoroughly. We present the global vision and advancement of UAV traffic management system, and then discuss the existing research trends. Understanding the complexity of future air transportation system, we discuss the necessary elements required for urban air traffic management system. We provide key strategic elements for an advanced urban traffic management through connectivity, open source, safety enhancement and automation to maximize its impacts on the innovative future air traffic and mobility system. In the end, we discuss the challenging security issues in urban traffic management system.

Absence of human in-charge in fully autonomous vehicles (FAVs) impose multi-pronged in-cabin monitoring in real-time to monitor any irregular situation. Moreover, occupant’s visual information (such as video, images) transmission from in-cabin of vehicle to the data center and control-room is required for various in-cabin monitoring tasks. However, this information transfer may cause a substantial threat to the occupants. Though applying some patch on face of the occupants protects personal information however, simultaneously it deteriorates the important facial information which is very crucial in monitoring various tasks such as emotion detection. Therefore, a human-computer-interaction (HCI) is required to manage this trade-off between facial information and facial anonymization without any important information loss. Also, this HCI based in-cabin monitoring system should be implicit to retain the users trust in the intelligent transportation system (ITS). In this paper, we have proposed generative adversarial network (GAN) based in-cabin monitoring approach for emotion detection in FAV. In this method, we proposed to generate an artificial (virtual) face having real-facial expressions to provide facial anonymity to the occupants while monitoring FAV cabin. Therefore, this method provides anonymous facial information which is essentially required in detection and monitoring tasks to avoid irregular situations. We have used both publicly available and our own in-cabin dataset to assess our proposed approach. Our experiments show satisfactory results.

Recent progress in parallel computing machines, deep neural networks, and training techniques have contributed to the significant advances in artificial intelligence (AI) with respect to tasks such as object classification, speech recognition, and natural language processing. The development of such deep learning-based techniques has enabled AI-based networks to outperform humans in the recognition of objects in images. The graphics processing unit (GPU) has been the primary component used for parallel computing during the inference and training phases of deep neural networks. In this study, we perform training using a desktop or a server with one or more GPUs and inference using hardware accelerators on embedded devices. Performance, power consumption, and requirements of embedded system present major hindrances to the application of deep neural network-based systems using embedded controllers such as drones, AI speakers, and autonomous vehicles. In particular, power consumption of a commercial GPU commonly surpasses the power budget of a stand-alone embedded system. To reduce the power consumption of hardware accelerators, reductions in the precision of input data and hardware weight have become popular topics of research in this field. However, precision and accuracy share a trade-off relationship. Therefore, it is essential to optimize precision in a manner that does not degrade the accuracy of the inference process. In this context, the primary issues faced by hardware accelerators are loss of accuracy and high power consumption.

Advances in the architecture of inference accelerators and quantization techniques of neural networks allow effective on-device inference in embedded devices. Privacy issues for user data, as well as increasing needs of user-specific services, have led to a need for on-device training. The dot product operation required in backpropagation can be computed efficiently by multiplier–accumulators (MACs) in the inference accelerator if forward and backward propagation of the neural network have the same precision. This chapter introduces a quantization technique to enable computation by the digital neuron inference accelerator with the same precision as that using the forward path. Updating the 5-bit weights with gradients of higher precision is challenging. To address this issue, this chapter also introduces a probabilistic weight update. It also describes the hardware implementation of the probabilistic weight-update scheme. The proposed training technique achieves 98.15% recognition accuracy on the MNIST dataset.

We propose a novel type of near-infrared (NIR) based sensory system for the detection of skin deformation induced by facial expressions. The proposed system utilizes the diffusion property which was called spatially resolved diffuse reflectance of NIR light in human skin as clues of detecting skin deformations. The sensor was mainly composed of NIR laser diode and a camera as a light source and receiver. To refine the information from captured IR raw images, we exploited principal component analysis based preprocessing procedure. By producing sensor data collecting experiment on several participants, we've verified the facial deformation tracking performance of the proposed sensory system.