Biswanath Samanta

Cloud robotics: Formation control of a multi robot system utilizing cloud infrastructure

The pathway for the concept of cloud robotics is continually unfolding and revealing new opportunities in science. With this, the focus of the research paper is aimed at identifying the progress completed towards the development of a full scale cloud infrastructure to implement formation control on a multi robot system. A small scale cloud infrastructure was developed utilizing a single virtual machine operating with the boundaries of a hypervisors resource pool. A robot with minimal hardware was constructed to work within the control of the cloud. Once the proof of concept on a lower tier has been completed, more advance robotics concepts, such as Null-Spaced-base behavior control and advanced neural network control, will be tested by offloading the computational load to the cloud infrastructure. The goal is to demonstrate the ability to simplify the robot hardware and implement control on a global scale utilizing the cloud infrastructure.

Development of a Network Infrastructure for Heterogeneous Robot and Control Systems Interactions

The use of robots in complex tasks such as search and rescue operations is becoming more and more common. These robots often work independently with no cooperation with other robots or control software, and are very limited in their ability to perform dynamic tasks and interact with both humans and other robots. To this end, a system must be developed to facilitate the cooperation of heterogeneous robots to complete complex tasks. To model and study human-robot and robot-robot interactions in a multi-system environment, a robust network infrastructure must be implemented to support the broad nature of these studies. The work presented here details the creation of a cloud-based infrastructure designed to support the introduction and implementation of multiple heterogeneous robots to the environment utilizing the Robot Operating System (ROS). Implemented robots include both ground-based (e.g. Turtlebot) and air-based (e.g Parrot ARDrone2.0) systems. Additional hardware is also implemented, such as embedded vision systems, host computers to support virtual machines for software implementation, and machines with graphics processing units (GPUs) for additional computational resources. Control software for the robots is implemented in the system with complexities ranging from simple teleoperation to skeletal tracking and neural network simulators. A robust integration of multiple heterogeneous components, including both hardware and software, is achieved.Copyright

Noise and Vibration Issues of Wind Turbines and Their Impact – A Review

Wind power is a rapidly growing technology, with an estimated 35% of national end-use electricity demand to be met from wind by 2050 in the US. With such a projected rapid growth, it is necessary to improve and innovate relevant technological areas with due considerations of possible impacts of wind energy harnessing through wind turbines. This paper presents a systematic review of current literature on the issues of noise and vibration of wind turbines and their impact on human health and wild life. The paper reviews the literature on the issues of noise and vibration in wind turbines, the generation mechanisms, the propagation, the impact on human health and wild life. The current status of technology and future developments to mitigate the health and environmental impacts of wind turbine noise and vibration are also reviewed. The paper includes a review of current standards on measurement of acoustic noise of wind turbines and data analysis.

Human Identification for Human-Robot Interactions

In co-robotics applications, the robots must identify human partners and recognize their status in dynamic interactions for enhanced acceptance and effectiveness as socially interactive agents. Using the data from depth cameras, people can be identified from a person’s skeletal information. This paper presents the implementation of a human identification algorithm using a depth camera (Carmine from PrimeSense), an open-source middleware (NITE from OpenNI) with the Java-based Processing language and an Arduino microcontroller. This implementation and communication sets a framework for future applications of human-robot interactions. Based on the movements of the individual in the depth sensor’s field of view, the program can be set to track a human skeleton or the closest pixel in the image. Joint locations in the tracked human can be isolated for specific usage by the program. Joints include the head, torso, shoulders, elbows, hands, knees and feet. Logic and calibration techniques were used to create systems such as a facial tracking pan and tilt servomotor mechanism. The control system presented here sets groundwork for future implementation into student built animatronic figures and mobile robot platforms such as Turtlebot.Copyright

Gesture Recognition for Control in Human-Robot Interactions

In co-robotics applications, the robots must be capable of taking inputs from human partners in different forms, including both static and sequential hand gestures, in dynamic interactions for enhanced effectiveness as socially assistive agents. This paper presents the development of a gesture recognition algorithm for control of robots. The algorithm focuses on the detection of skin colors using monocular vision of a moving robot base where the inherent instability negates the effectiveness of methods like background subtraction. The algorithm is implemented in the open-source, open-access robotics software framework of Robot Operating System (ROS). The video feed from the camera is converted into several color spaces, including RGB and YCbCr. Pixels observed in the raw video feed as skin are randomly selected and their properties in each of the color spaces are recorded. A cylinder of infinite length is constructed out of the best fit line for both color spaces, and all points lying within both cylinders are accepted as a skin tone. The gesture recognition features are extracted from the filtered image and can be used for planning the motion of the robot. The procedure is illustrated using the on-board camera on an unmanned aerial vehicle (UAV).Copyright

Neuromodulation based control of an autonomous robot

The paper presents a control approach based on vertebrate neuromodulation and its implementation on an autonomous mobile robot platform. A neural network is used to model the neuromodulatory function for generating context based behavioral responses to sensory signals. The implementation of the neuronal model on a relatively simple autonomous robot illustrates its interesting behavior adapting to changes in the environment.

Control of Autonomous Robots Using the Principles of Neuromodulation

The paper presents a control approach based on vertebrate neuromodulation and its implementation on an autonomous robot platform. A simple neural network is used to model the neuromodulatory function for generating context based behavioral responses to sensory signals. The neural network incorporates three types of neurons — cholinergic and noradrenergic (ACh/NE) neurons for attention focusing and action selection, dopaminergic (DA) neurons for curiosity-seeking, and serotonergic (5-HT) neurons for risk aversion behavior. The implementation of the neuronal model on a relatively simple autonomous robot illustrates its interesting behavior adapting to changes in the environment. The integration of neuromodulation based robots in the study of human-robot interaction would be worth considering in future.Copyright

Network performance of wireless cloud-based robots with local processing

With the impending nature of the Internet of Things (IoT), it is important that devices proposed to be networked to the system operate in such a way as to be non-disruptive to the operation of the network. Devices should only use network bandwidth as needed for their operation. To maintain a minimal network footprint which is conducive to the operation of the network, IoT devices should perform a significant amount of processing locally and not depend on remote or cloud resources any more than necessary. This work presents development of a cloud computing infrastructure for networked heterogeneous robotic systems in open-source robot operating system (ROS). This work demonstrates the minimal impact on network performance by devices which use a significant amount of local processing for their operation. Using a virtual datacenter with (5) host servers of 68 GHz processing capability and 160 GB RAM, a number of Kobuki Turtlebots and LEGO EV3 robots are connected to cloud services on the network through the Robot Operating System. Each robot is connected to the cloud via wireless network. A varying number of Turtlebots are tested under low- and high-bandwidth conditions while performing computations related to robot operation locally. The latency and data integrity of the network connections are measured under these conditions and presented along with recommendations for further work.

Empirical mode decomposition of EEG signals for brain computer interface

Motor imagery (MI) based brain-computer interface (BCI) systems show potential applications in neural rehabilitation. In MI-BCI systems, the brain signals from movement imagination, without actual movement of limbs, can be acquired, processed and characterized to translate into actionable signals that can be used to activate external devices. However, success of such MI-BCI systems, depends on the reliable processing of the noisy, non-linear, and non-stationary brain activity signals for extraction of characteristic features for effective classification of MI activity and translation into corresponding actions. In this work, a signal processing technique, namely, empirical mode decomposition (EMD), has been proposed for processing EEG signals acquired from volunteer subjects for characterizing MI activities and activity identification.

Development of a Mechatronics Course Integrated With Lab

This paper reports the development of an introductory mechatronics course in Mechanical Engineering (ME) undergraduate program at Georgia Southern University. This an updated version of an existing required course in the ABET accredited BSME program. The course covers three broad areas: mechatronic instrumentation, computer based data acquisition and analysis, and microcontroller programming and interfacing. This is a required 3-credit course in the ME program with updated computing application specific content reinforcing theoretical foundation with hands-on learning activities of the existing course. The course has four contact hours per week with two hours of lecture and two hours of interactive session of problem solving and laboratory experiment. For each topic covered, students get the theoretical background and the hands-on experience in the laboratory setting. Both formative and summative assessment of the students’ performance in the course are planned. Both direct and indirect forms of assessment are considered. The paper reports the details of the course materials.Copyright