Cognitive Computing for Human-Robot Interaction: Principles and Practices

By Mamta Mittal

Cognitive Computing for Human-Robot Interaction: Principles and Practices explores the efforts that should ultimately enable society to take advantage of the often-heralded potential of robots to provide economical and sustainable computing applications.

This book discusses each of these applications, presents working implementations, and combines coherent and original deliberative architecture for human–robot interactions (HRI). Supported by experimental results, it shows how explicit knowledge management promises to be instrumental in building richer and more natural HRI, by pushing for pervasive, human-level semantics within the robot's deliberative system for sustainable computing applications.

This book will be of special interest to academics, postgraduate students, and researchers working in the area of artificial intelligence and machine learning.

Key features:

• Introduces several new contributions to the representation and management of humans in autonomous robotic systems;
• Explores the potential of cognitive computing, robots, and HRI to generate a deeper understanding and to provide a better contribution from robots to society;
• Engages with the potential repercussions of cognitive computing and HRI in the real world.

• Introduces several new contributions to the representation and management of humans in an autonomous robotic system
• Explores cognitive computing, robots and HRI, presenting a more in-depth understanding to make robots better for society
• Gives a challenging approach to those several repercussions of cognitive computing and HRI in the actual global scenario

• Language: English
• Publisher: Academic Press
• Release date: Aug 13, 2021
• ISBN: 9780323856478

Book preview

Preface

Mamta Mittal¹, Rajiv Ratn Shah² and Sudipta Roy³, ¹Department of CSE, G. B. Pant Government Engineering College, New Delhi, India, ²Department of Computer Science and Engineering (joint appointment with the Department of Human-centered Design), IIIT-Delhi, New Delhi, India, ³PRTTL, Washington University in Saint Louis, Saint Louis, MO, United States

Cognitive computing is the use of computerized models to simulate the human thought process in complex situations where the answers may be ambiguous and uncertain. Using self-learning algorithms that use data mining, pattern recognition, and natural language processing, the computer can mimic the way the human brain works. So it can be incorporated with Human–Robot Interaction (HRI) for assisting humans in various forms.

The study of interactions between humans and robots is thus fundamental to ensure the development of robotics and to devise robots capable of socially interacting intuitively and easily through speech, gestures, and facial expressions. HRI is an integrative field that has provided considerable improvement in various applications, including human–computer interaction (HCI), robotics and artificial intelligence, the service robots combined with cognitive computing to design intelligent robotic services. This robotics provides novel services in terms of companionship for the elderly and disabled and does household chores in the home environment. It lays at the crossroad of many subdomains of AI and, for effect, it calls for their integration: modeling humans and human cognition; acquiring, representing, manipulating in a tractable way abstract knowledge at the human level; reasoning on this knowledge to make decisions; and eventually instantiating those decisions into physical actions both legible to and in coordination with humans. Unique features of Cognitive Computing for Human–Robot Interaction: Principles and Practices are as follows:

• Introduces several new contributions related to the representation and the management of humans in an autonomous robotic system.

• The main strength of this powerful and reputable collection is to collect and provide knowledge regarding cognitive computing and HRI.

• It will give a challenging approach of those several repercussions of cognitive computing and HRI in the actual global scenario.

• This covers a wide spectrum of topics that will be of interest among groups of readers consisting of professionals, educator, academics, and students.

• It will also help educators be aware of the changes and of what society needs from them.

Supported by experimental results, this book focuses on how explicit knowledge management, both symbolic and geometric, proves to be instrumental to richer and more natural HRIs by pushing for pervasive, human-level semantics within the robot’s deliberative system. It has been organized into 17 chapters.

Chapter 1, Introduction to Cognitive Computing and its Various Applications, explains that cognitive computing is an intelligent system that converses with and mimics the human being in a natural form by learning at scale, reasoning with purpose. It falls under the third era of computing and now has attracted considerable attention in both academia and industry. Nowadays, there is explosive data growth, business conditions are also changing rapidly; so smart, hassle-free, and enhanced interactions amongst human beings and technology can be effectively addressed by cognitive systems. Some examples of personal assistants that use cognitive computing are Alexa, Siri, Google Assistant, and Cortana. Some more applications that can use cognitive computing to gain benefits from this type of technology are cognitive computing for changing business values, financial and investment firms, healthcare and veterinary medicine, travel and tourism, health and wellness, education and learning, agriculture, communication, and network technology.

Chapter 2, Recent Trends Toward Cognitive Science: From Robots to Humanoids, focuses on how cognitive processing is to replicate the human thought processes in a computerized model. Employing self-learning algorithms that use data mining, pattern recognition, and natural language processing concerning this specific subject in which the computer can mimic the human brain function. Following the diversity of user’s interpretation of problems, the cognitive computing system presents the reorganization of types of data and sways meaning and analysis. When concluding, the cognitive computing system usually relies on considering contradictory proof and proposes a solution that is best more than precisely right. The systems with cognitive computing ability create situations quantifiable. Machines with cognitive systems recognize as well as excerpt factor characteristics for example location, time, history, work, or description to portray the dataset suitable for a person or a relying implementation involved in a particular method at an exact schedule and location. These systems reframe the scenery involving the association of individuals and their progressively all-encompassing digital environment. This has been shown to play a larger role as the mentor or assistant for the user; similarly, they may behave digitally independently for resolving various problems.

Chapter 3, Cognitive Computing in Human Activity Recognition With a Focus on Healthcare, presents that human activity recognition is a quintessence to empower a robot to distinguish the conduct of a personal care-receiver. As opposed to outward appearances, an activity recognition can see practices of a consideration beneficiary, who might be a senior adult, a youngster, or a chronic patient. Through human activity recognition, a robot tracks the care patient activity and perceives human practices like unhealthy habits and anomalous activities. However, patient activity recognition through simple images is a highly challenging task. Several challenges such as the likeness of unmistakable human behaviors, disorder background, similarities in different human activities may significantly reduce the classification performance. Because of rapid developments in cutting-edge machine learning models, substantial solutions can arise from distinct deep learning algorithms, including Convolutional Neural Network (CNN), Generative Adversarial Network. In this chapter, the authors review several cognitive computing approaches in the advancement of HRI, especially in healthcare industries.

Chapter 4, Deep Learning-based Cognitive State Prediction Analysis Using Brain Wave Signal, presents the analysis of cognitive state during different learning tasks using EEG signals. Online learning tools such as video-conferencing, multimedia lessons, digital materials, and e-learning platforms with options for both real-time learning and self-paced learning provide a pleasant and immersive experience. In addition to these features, assessment of cognitive state during the learning phase has been proven to improve the efficiency of online learning. The analysis of cognitive state during different learning tasks using EEG signals that were obtained using 128 channel Emotive Epoch headset device is the main focus of this study. Artifacts prominent in raw signals were filtered by using linear filtering. To determine the exact concentration levels, the fuzzy fractal dimension measures and the Discrete Wavelet Transform were adapted to the same extracted Electroencephalogram (EEG) signals for feature extraction. The extracted parameters are then classified into concentration levels using the deep learning algorithm Enhanced Convolutional Neural Network (ECNN), which has proven to be of higher accuracy compared to other classifiers. ECNN can then be used to control cognitive states as a feedback mechanism.

Chapter 5, EEG-Based Cognitive Performance Evaluation for Mental Arithmetic Task, discusses an appropriate framework to assess participant’s cognitive performance based on their brain activity dynamics recorded through an EEG device. To this aim, the authors have used a publicly available EEG dataset. The dataset contains EEG recording of 36 subjects before and during a mental arithmetic task. The participants were divided into two subgroups (good and bad performers) based on the accuracy of the task performed. A simple but novel approach has been proposed to summarize these window-level features and formulate the signal-level descriptor. The descriptor thus formed, captures the distribution of the feature values effectively. Experimental results suggest that the proposed descriptor, obtained after summarizing the window level EEG domain features, performs satisfactorily in discriminating between the two sets of performers. Mean classification accuracy obtained was about 85% using Gaussian naïve Bayes classifier which outperformed EEG domain feature-based classification models.

Chapter 6, Trust or No Trust in Chatbots: A Dilemma of Millennial, empirically investigates what dilemma millennials have regarding the perception of trust while interacting with chatbots and how the perception of trust can be measured in terms of this research context. The major findings of the study are that millennials’ perception of trust has been measured into two dimensions. The first dimension of trust is cognitive-based trust which represented that we choose whom we will trust in which respect and under what circumstances, and we base the choice on what we take to be good reasons, constituting evidence of trust-worthiness conversing with the chatbot. The second dimension is affect-based trust which is made by emotional bonds between individuals that go beyond a regular business or professional relationship. So, it is recommended for the organizations to implement chatbots in their premises for the initial interaction with the millennials.

Chapter 7, Cognitive Computing in Autonomous Vehicles, discusses the neuromorphic architecture and how it is inspired by the human brain which mimics micro neurobiological architecture present in nervous systems, and the Von Neumann architecture model combines to form what we call Cognitive Computing, its applications and advantages in AVs. For this, It takes into account hardware components and mathematical models required for the design of an autonomous vehicle. It focuses on different Cognitive Artificial Intelligence techniques and Algorithms that will help us to achieve closeness to human-level performance or what we call Level 5 autonomy in AVs having unlimited Operational Design Domain. Achieving level 5 autonomy is an extremely difficult task because it requires almost perfect decision making, Object and Event Detection and Response, localization even in uncertain conditions like cloudy weather, fog, extreme darkness, and rain, which act as a forestall to vision task and localization. That is where cognitive computing comes into the picture. Cognitive computation techniques enhance the accuracy of the model to achieve human-like performance in decision making or even in object detection. Yet even above all this, there is greater scope for development, as perfect Level 5 autonomy is still not achieved. Furthermore, this chapter explores the further developments possible in this field of AVs and the effects which it will cast on future generations.

Chapter 8, Optimized Navigation Using Deep Learning Technique for Automatic Guided Vehicle, presents that autonomous driving has passed the point of being called the biggest step, as the smart car revolution is already taking shape around the world. Self-driving cars are relevant if not prevalent and the biggest obstacles to reach mass adoption are customer acceptance, cost, infrastructure, and the reliance on several onerous algorithms that include perception, lane marking detection, path planning, and variation in pathways. This study tackled the mentioned problems with a straightforward and cost-effective solution, using end-to-end learning and replacing the numerous sensors with a camera and commandeering just the forward, backward, left, and right controls. In this research, the authors have used the most popular method of deep learning that is CNN to train the collected data on the VGG16 model. Later these have optimized directly by the proposed system with cropping each unnecessary image and mapping pixels from a single front-facing camera to direct steering instructions. It has been observed from the experimental work that the proposed model has given a better result than the existing work that is increasing in the accuracy from 88% (Udacity training dataset) to 98% (proposed). This model is suitable for industrial use and robust in real-time scenarios, therefore, can be applied in modern industrialized systems.

Chapter 9, Vehicular Middleware and Heuristic Approaches for ITS Systems of Smart Cities, discusses that a smart city comprises the intellectual use of the vehicular system. This technical revolution from wireless telephony to Vanet networks provides Intelligent transport system for Smart cities. The middleware and heuristic approaches of Vehicular system are essential for designing and analyzing research problems related to smart cities. The review of the literature played a critical part in determining numerical knowledge about the smart cities with the vehicular system, middleware and heuristic approaches, find gaps in published research, and produce new original ideas for the intelligent transport system of Smart cities.

Chapter 10, Error Traceability and Error Prediction Using Machine Learning Techniques to Improve the Quality of Vehicle Modeling in Computer-Aided Engineering discusses how machine learning techniques playing the role of bug prediction and effectively increases the accuracy rate in vehicle modeling. While designing a vehicle, the designer follows a step-by-step process to reduce the unintended software behavior to complete the model. Locating Errors in industry-size software systems is time-consuming and challenging. Hence an automated approach is proposed to trace the errors helps the CAD engineer to complete the designing task in less time. Error Traceability and Error Prediction are the two major tasks focused on the designing stage. Software fault prediction techniques are used to predict the faults in the software and machine learning techniques is playing an important role in detecting the software default. Bug prediction and correction of bugs improve the software quality and reduces the maintenance cost.

Chapter 11, All About Human–Robot Interaction, illustrates a generalized framework and metric taxonomies for robot design through a detailed review of the literature. The framework presents an end-to-end product design overview for hobbyists and researchers interested in the general steps to create a fully autonomous, consumer-focused social robots with perception and reasoning capabilities that aim to achieve high technology readiness. The latest developments in science and technology have led to the expansion and deployment of robots in various applications covering all spheres of human life. This necessitates the interaction of humans with robots on a larger scale, as past evidence suggests the robots are considered more as companions rather than machines/tools. HCI provides insights in understanding and improving interactions with computer-based technologies. However, HRI takes a cue from HCI by introducing autonomy, physical proximity, and the ability to make decisions in addition to HCI techniques for a robotic system, which makes HRI a distinct area for research. Taking into consideration the huge number of interactions between humans and robots, there is a stringent requirement to standardize and make fixed protocols to ensure the usage of robotic technology in a responsible and principled way.

Chapter 12, Teleportation of Human Body Kinematics for a Tangible Humanoid Robot Control, presents that People learn best when they use sensory-based perceptual learning styles. To model this action-learning, a didactic design was used to create an instructional resource and applied through humanoid robot interaction. The quasiexperimental result of the interaction analysis has revealed higher retention of learning contents by participants. This work is a systematic approach to involve the kinetic movement of human limbs with sensory organs with teleportation of gesture movements on humanoid robots, which has made twofold coordination between human and machine in a live interaction. People with learning difficulty as well as gamification of learning in elementary classes can be addressed with this augmented approach in conventional pedagogy. The algorithmic percept action sequence has created a unified order of closed-loop interaction across different cognitive level people. The experimental results have revealed substantial evidence of learning enhancement and higher order logical understanding by using the proposed immersive extension of natural body movement identification by machines and teleportation of the same to another machine.

Chapter 13, Recognition of Trivial Humanoid Group Event Using Clustering and Higher Order Local Auto-correlation Techniques presents that a video surveillance system is explicitly established to manage a small human group. The proposed methodology concentrated on the trivial humanoid groups that remained in an identical location for some time and characterized the group activity. The defined methodology has widespread applications in numerous areas such as video reconnaissance systems, cluster interface, and activities classification. The video surveillance system primarily covenants with the action recognition and classifying the cluster activity by considering violent activities such as fighting. The steps in action recognition include generation of frames, segmentation using fuzzy c-mean clustering, feature extraction by completed local binary pattern and high order local autocorrelation, classification by Recurrent Neural Network. Spatial Gray Level Difference Method extracts the statistical features while bag-of-words technique creates vocabulary features set which is derived from Local Group Activity descriptors. CNN classifier employed to classify the human activities.

Chapter 14, Understanding the Hand Gesture Command to Visual Attention Model for Mobile Robot Navigation: Service Robots in Domestic Environment, presents that in recent years, robotic systems and techniques have acquired the unprecedented capacity for perceiving and understanding their world not just in a low-level manner but even close to humanly understandable concepts. HRI is used frequently in the care of the aged and the disabled population. Human behavior is expected to achieve natural interaction from these robots. Human activity is essential both before and after contact with a human user is initiated. Intelligent service robots in evolving fields of robotic technologies, from entertainment to healthcare, are currently being built to satisfy demand. The service robots have controlled by nonexpert users, and direct contact with them and their human users will be their support activities. With these service robotics, human-friendly social features are typically favored. Individuals tend to use voice commands, responses, and suggestions to express their peers’ opinions.

Chapter 15, Mobile Robot for Air Quality Monitoring of Landfilling Sites using Internet of Things, discusses air quality monitoring is a vital mechanism for continuously observing the quality of air in the environment. Generally, landfilling sites are filled with mixed waste of various materials, this mixed waste generates stinky gases. Robots are the solution for monitoring the air quality of landfilling sites. In this study, an IoT and cloud server–enabled mobile robot–based model has been proposed for monitoring the air quality at the landfill sites. LoRa radio-based sensor nodes are embedding with distinct gas sensors for sensing in landfill sites and transmits to the mobile robot. A mobile robot is an integration of LoRa radio and GPRS communication, it transmits the sensory data to the cloud server via internet protocol. The sensory data of different gases in the landfilling sites are recorded in the cloud server and also a graphical representation of the sensory data is discussed briefly.

Chapter 16, AI and IoT Readiness: Inclination for Hotels to Support a Sustainable Environment, presents that Smart Cities has been one of the key driving factors for the urban transformation to a low carbon climate, sustainable economy and mobility in recent years because of the alarming situation of Global warming. One of the industries with swift growth is the hotel sector and hence is one of the key contributors to carbon emission and leaves environmental footprints. The new emerging concept of sustainable tourism is envisaged as an important part of the Smart Cities paradigm. Improving sustainability by saving energy is becoming a primary task today for many hotels. A great opportunity is provided by AI and IoT to assimilate different systems on a platform by encouraging and assisting hotel guests to operate through a single device and optimizing hotel operations. Current research focuses to identify the strategic positions of a hotel in terms of sustainability, AI and IoT technology. Components that will be considered by Hotels for the strategic intention of adopting AI and IoT for environmental sustainability. Different development and modification needed to be taken if management wants high sustainability readiness and/or IoT readiness. This conceptual paper constructs the comprehensive study and systematic review of different areas where the Hotels can feasibly implement AI and IoT for improving sustainably.

Chapter 17, Design and Fabrication of an Automatic Classifying Smart Trash Bin Based on IoT, presents that Municipal Solid Waste is an increasing waste resource and challenging task to solve environmental pollution and waste collection problems. The smart trash bin is sufficient equipment to reduce human intervention and improve living conditions, lacking multifunctional purposes. To solve the problem that traditional trash can lacks self-identification of recyclable garbage and nonrecyclable garbage, a smart trash bin based on WIFI environment and Arduino control is proposed. The smart trash bin takes Arduino as its master controller and interacts with Arduino in the WIFI environment to realize automatic garbage classification and communication between people and the trash bin. This paper aims to describe the hardware and software design ideas of trash bins in the implementation process, and the mechanism of garbage classification and the application of the Internet of Things in garbage recycling are emphatically studied. The results indicate that the system can run stably and achieve accurate classification within 2 seconds, which reduces the waste of resources, changes the traditional garbage management mode, improves management efficiency, and realizes recycling of garbage.

The editors are very thankful to Rachel Pomery, Editorial Project Manager: Elsevier and Series Editor Professor Arun Kumar Sangaiah for providing us the opportunity to edit this book.

Chapter 1

Introduction to cognitive computing and its various applications

Sushila Aghav-Palwe and Anita Gunjal,    School of CET, MIT World Peace University, Pune, India

Abstract

Cognitive computing is an intelligent system that converses with and mimics the human being in a natural form by learning at scale, reasoning with purpose. Cognitive computing is the third era of computing and now cognitive computing has attracted considerable attention in both academia and industry. Machines and humans’ intelligence gets combined to solve the most complex problems of the world. Complicated problems can be solved by computing framework without intervention of humans. Natural language processing with emotion analysis, artificial intelligence (AI), machine learning, neural networks are building blocks of cognitive computing process to tackle problems as the way human beings do. Nowadays, advance technologies adapt cognitive computing in many areas to assist human experts in smart decision making for the betterment of businesses. Current technology expectations are to make human life better and to help them work in better ways. Nowadays, there is explosive data growth, business conditions are also changing rapidly so intelligent, hassle-free, and enhanced interactions amongst human beings and technology can be effectively addressed by cognitive systems. AI is in use in many apps like the Alexa: Amazon voice assistant, Netflix and Amazon algorithms which recommend the next to watch or buy. Some examples of personal assistants that uses cognitive computing are Alexa, Siri, Google assistant, and Cortana. Advancement of technology and its adoption in the public and private sectors will greatly affect the future of cognitive computing due to technology evolutionary paths and trends. Cognitive systems must be adaptive, interactive, iterative, and stateful and contextual in commercial and widespread applications. Some of the applications that can use cognitive computing to gain benefits from this type of technology are cognitive computing for changing business values, Financial and Investment firms, Healthcare and veterinary medicine, Travel and Tourism, Health and wellness, Education and learning, Agriculture, Communication and network technology.

Keywords

Artificial intelligence; cognitive computing; contextual awareness; natural language processing; machine learning

Introduction

In this era of modern computing system, cognitive computing is emerging technological mode, as process automation is expected to evolve the old systems. According to Gartner, cognitive computing will change the technology era unlike none of the technology introduced in the last two decade. In technology, cognitive computing is becoming modern buzzword, capturing the attention of many entrepreneurs and tech enthusiasts.

Cognition is the combination of brain related processes and activities used to learn, perceive, think, understand and remember, to acquire knowledge and understand, experience, and the senses. Cognitive Science is the broad scientific study of the human-mind, the brain, and human-intelligent behavior. As cognitive system stimulates the mechanism of human thinking, the use-cases and benefits of cognitive computing are much more advanced and smarter than artificial intelligence (AI) systems. As it is able to analyze and handle Big data, cognitive computing is capable to be used in today’s smart and complex system to solve real-life problem. AI creates new ways to solve problems better than human but cognitive computing tries to replicate human behavior to solve problems (Megha et al., 2017). Cognitive system must adopt the cognitive skills of human like perception, decision making ability, motor skills, language skills and social skills to solve problems. AI can only be intelligent as how individual teach it, but it is not suitable for current cognitive era. Advanced cognitive computing framework uses natural language processing (NLP) with context-aware emotion intelligence, AI, machine learning, neural networks as a foundation to tackle complex routine problems as humans-being. Cognitive computing is defined as An advanced system that learns at scale, reason with purpose and interacts with humans in a natural form (Megha et al., 2017).

Cognitive system learns the patterns and advises human react appropriately depending on acquired knowledge. It provides an assistance to human rather than one completing the task. The main purpose of cognitive computing is to assists people in decision making which gives them superior accuracy in analysis. Since cognitive computing has the ability to analyze data faster and more accurately, there is no need to worry about the incorrect decisions. If we see example of healthcare system using AI and using cognitive computing, treatment decisions are made without consulting with human doctor in AI system, whereas cognitive computing work one step ahead and assist doctors for disease diagnosis with the help of data repositories and data analysis which helps in smart decision