Search

Robot learning by demonstration is a research paradigm that can play an important role in expanding areas where robots can be applied. An easy method to get a policy to reproduce the demonstrated behavior is to learn a model that maps directly from demonstrations to action. This is referred to as behavioral cloning (BC). In this paper, a novel robotic BC method for directly imitating a policy from trajectories generated by performing a task is proposed. The proposed method not only allows a user to obtain a demonstrated trajectory for performing a task by manipulating the robot, but also automatically generates a trajectory by having the robot perform the task on behalf of the user. Experiment results showed that the proposed method can effectively learn to make accurate behavior prediction for robot manipulators.

Recently, reinforcement learning utilizing deep learning, called deep reinforcement learning, is being developed for discontinuous action space. It is shown that deep reinforcement learning has higher playing skills than those of human in the video game as Atari. The development of deep reinforcement learning for discontinuous action space, has been extended to the continuous action space as robot control area. The area of control deals with issues closely related to real world environment. However, deep reinforcement learning has limitations to solve control problems in the real world due to numerous trial-and-error based training. To address the limitations of learning in the real world, virtual world environments are used to emulate control problem in real world. In this paper, to solve a complex robot manipulation task, the learning robot manipulation based modular reward shaping is proposed. Since it is difficult to solve the complex task using only single primitive manipulation skill with a reward, it is required to combine various primitive manipulation skills with various rewards. In simulation results, it is shown that the combination of the reward functions affects to the robot manipulation performance.

Advances in deep reinforcement learning enabled cost-efficient training of control policy of physical robot actions from robot simulators. Learning control policy in a simulated environment is cost-efficient over learning in a real environment. Reward engineering is one of the key components to train efficient control policy. For tasks with long horizons such as navigation and manipulation, a sparse reward is providing limited information. The robot simulator for a physical engine of physical robot manipulation has made it easy for researchers in the field of deep reinforcement learning to simulate complicated robot manipulation environments. In this paper, A robot manipulation simulator and a deep RL framework are utilized for implement a training control policy by utilizing previous experiences. For implementation, Recent innovation Hindsight Experience Replay (HER) algorithms with previous experiences to calculate dense rewards from a sparse reward is leveraged . Proposed implementation showed an approach to investigate the reward engineering method to formulate dense reward in robot manipulator tasks.

This paper provides an experimental study of reinforcement learning on IoT devices using distilled knowledge, whose a teacher with a well-trained model transfers to a student with a new model to be trained. The experimental results show that the distilled knowledge is effective to a new model training on IoT devices.

Developing an autonomous indoor mobile robot navigation system from scratch is very difficult and it takes a long time. It requires a series of complex tasks such as handling given sensor inputs, calculating all the information needed for autonomous navigation, and controlling actuators required for movement. In this paper, an autonomous navigation system for indoor mobile robots is introduced mainly based on open source provided by the robot operating system. The presented system is capable of autonomously navigating an unstructured indoor environment avoiding collision with static or dynamic objects. To this end, the system consists of three main modules: mapping, localization, and planning. The mapping module builds a global map for an unknown environment by means of a simultaneous localization and mapping algorithm based on laser scanner data. The localization module estimates the mobile robot’s pose within the prebuilt map by way of an adaptive Monte Carlo localization approach. The planning module builds a local cost map for collision avoidance, generates collision-free trajectories to reach a goal pose based on the cost map, and produces control commands to follow the trajectories. The presented system has been tested not only in simulation environments built in the Gazebo simulator but also in real environments utilizing the Jackal mobile robot, to validate its performance for autonomous navigation including collision avoidance.

Today, a paradigm in the manufacturing industry has been shifted from mass production to small and custom production. To adapt to today’s industrial expectations, this paper presents position-based visual servoing of multiple robotic manipulators. The presented visual servoing consists of two modules: task planning and motion planning. The task planning module manipulates each robotic arm not to collide with each other by distributing a task to them in a systematic manner. The motion planning module employs the MoveIt framework to solve inverse kinematics for each robotic arm so that it moves each end-effector’s pose to the desired pose to complete the task. In addition, the module controls a robotic hand to grasp or release a target object. The presented visual servoing has been tested and verified in the Gazebo simulator.

Applying deep learning to symbolic planning is not straightforward because they differently internalize knowledge. One represents knowledge numerically while the other does symbolically. Hybrid methods that take strengths from both techniques will deliver better performance on automated planning. This paper presents a method to improve the performance of symbolic planning by applying LSTM recurrent networks to inferring the guidance knowledge for search tree node expansion. The networks learn sequential patterns across previous node expansions and generalizes the learned knowledge to predict promising actions and their probability scores at the present expansion step. The prediction consists with node cost estimation for heuristic search. A simple experiment vindicates the method of this research.

This paper addresses a continual lifelong learning problem that learns incremental multiple tasks in real-world environments. We overview and summarize representative approaches and categorization of the state-of-the-art in continual learning. Comparable scenarios, benchmark datasets, and baseline approaches for different continual scenarios introduced in this paper. We suggested a comparison of the differences and similarities with other machine learning methods. We also report real-world applications, especially robots and healthcare fields. We summarize current states and suggest future direction of continual learning problems.

This paper proposes a system design for automation in multi-agent-based manufacturing systems to conduct a given complex task automatically by controlling multiple robotic manipulators in a systematic manner. To this end, the proposed system is designed with three-module configurations: environmental perception, task planning, and motion planning. The environmental perception module utilizes a vision sensor to recognize all objects placed on the workspace and extract their unique ID, size, and pose. The task planning module divides a given task into primitive skill levels and distributes each primitive skill to the associated robotic manipulator with the relevant object information in a systematic manner for robotic manipulators not to collide with each other. The motion planning module determines the motion of a robotic arm and a robotic hand by solving inverse kinematics for the robotic arm and by opening or closing two fingers. The proposed system has been tested and verified in real robot environments through a complex task "peg in hole" that requires at least two robotic manipulators.

Non-Hermitian singularities known as exceptional-points (EPs) have been shown to exhibit increased sensitivities but the observation of EPs has so far been limited to wavelength scaled systems subject to diffraction limit. We propose a novel approach to EPs and report their first observation in plasmonics at room temperature. The plasmonic EPs are based on the hybridization of resonances in multilayered plasmonic crystals to reach a critical complex coupling rate between nanoantennas arrays, and, resulting in the simultaneous coalescence of the resonances and loss rates. Because plasmons shrink the wavelength of light to make it compatible with biological relevant substances, enhanced sensing of anti-Immunoglobulin G, the most common antibody found in blood circulation, is observed at attomolar concentration. Our work opens the way to novel class of nanoscale devices, sensors, and imagers based on topological polaritonic effects.

This paper introduces a smart sleep care system based on IoT bed sensors and IoT LED. Quality sleep is influenced by sleep duration as well as bio-rhythm based sleep pattern. IoT bed sensor acquires body pressure on the bed and the sensor detects user's pose and biosignal patterns; respiration and pulse. IoT LED controls brightness and light temperature to induce sleep (dimming without blue) and wake up (brightening with blue). To improve quality of sleep, this system induces sleep time using LED control to support sufficient sleep duration and analyzes sleep pattern from the biosignal to wake up refreshed.

This paper describes crowd-sourcing home energy efficiency measurement system that includes ways for logging, analyzing, mining, measuring and recommending energy usage activity at home. These activities are based on HVAC devices (air-conditioner, boiler, and temperature sensors. Our system can help to build more valuable applications for management, resource optimization and upgrading user experience. In energy efficiency analyzing, heterogeneous environments make it to be difficult to accurately analyze and predict energy use in mild climates than in more severe climate. Individual sensing and measuring can cause inaccurate data and each home's analysis does not make a respectable result. In this paper, we can achieve crowd-sourcing home data gathering and manipulating data analysis for energy efficiency at home.

Nowadays, much research in recent years has focused on IoT (Internet of Things). The home domain is the most important research area of IoT, because there expected accounts of home smart device for over 40 percent of connected device excepting mobile phone. Furthermore, enormous data is generated by home smart device. There is growing concern, but the previous works didn't address enough to manage and analyze home data. The purpose of this study is to describe and examine to manage the aggregated home IoT data based on SWO (Smart home Web of Objects), and SWO analytics platform. We shows the implementation of SWO analytics platform and a case study using real data from smart metering devices for analysis of appliance usage patterns.

We report the nanogap resonator with the straight waveguide without the ring shape resonator, which is replaced with a straight waveguide, metallic layer, and nanogap. We use the numerical method to investigate the resonant properties of the structure. The results reveal that the proposed structure has the band stop characteristic.

This paper describes user activity recognition in home network - several ways for developers to define some model for logging, analyzing, mining, measuring and recommending user's activity. These activities are based on home networking between devices (gadgets, sensors, appliances, and some software applications at computers) and providing some standard interfaces for these activities. Users spend a lot of time at home and even if mobile gadget should be synced within home appliances. User's life pattern can be achieved from some data such as network traffic, control data, energy consumption, pims (calendar, email, todo, and so on), contents and actual user's profile. In this paper, we will explain how to care a diabetic patient using user activity recognition at home.

This paper proposes an efficient autonomous fault diagnosis system for connected home. The system takes a model-based approach and initiates an automated fault diagnosis without user intervention when an error is notified from the home network resource management system. Our system performs two fault diagnosis methods of resource probing and code mapping to find the root cause of the error in the resource relation model. Our experiments show that it can identify a smallest set of symptoms effectively and efficiently.

We demonstrated simultaneous RGB light generation in z-cut MgO-doped periodically-poled LiNbO 3 crystal with ridge waveguide. The green (532nm) and blue (460nm) lights were generated by quasi-phase-matched (QPM) second-harmonic generation pumped at 1064nm and 920nm. Various visible laser colors were obtained by adjusting coupled pump source and generated visible light intensity ratio.

We fabricated QPM-SHG (quasi-Phase-Matched Second-Harmonic-Generation) ridge type waveguide using PPMgLN Dry-etch Fabricaion process. We measured sidewall Roughness of ridge waveguide using AFM. We calculated optical scattering Loss with sidewall roughness RMS and Ridge Waveguide structure inspection result. As a result of calculated loss, sidewall scattering roughness little contributes to the propagation loss. By using optimized period and NLD dry-etch Process, high conversion efficiency can be obtained by using a SHG device at room temperature.

This paper presents a virtual energy planning(VEP) system that monitors energy usage of home appliances and simulates it when one or more virtual devices are removed or newly added. With the system users could minimize energy waste and manage usage pattern efficiently. 1

In order to accommodate different requirements of reliable multicast applications, the real-time reliable multicasting support framework must possess generic yet rich features for detecting and reporting message losses with tight latency bounds. One highly promising concrete formulation of a multicast framework is the Real-time Multicast & Memory Replication Channel (RMMC) scheme. In this paper, we present the Delay-Bounded Reliable RMMC (DBR-RMMC) scheme, which extends the RMMC scheme in order to yield tight bounds on the latencies in detecting and reporting message losses over RMMCs and to enable application layer to initiate timely recovery actions. The DBR-RMMC scheme offers an API that allows application developers to freely explore the various design dimensions of reliable multicast applications while assuring timeliness. We also discuss a middleware subsystem devised to support DBRRMMC, followed by a summary of several analytical results on the bounds for latencies in reacting to the occurrence of a message loss. Experiments involving a real-time video streaming application have been conducted and encouraging results are presented.

Network-based real-time computing applications which require tightly bounded end-to-end delays have been increasing at a steady rate in recent years. A messaging layer that yields tight bounds on the latencies in detecting message losses and enables the application layer to initiate a timely recovery action has thus become highly desirable. In this paper we present a new reliable real-time messaging scheme called the Delay-Bounded Reliable Messaging Scheme (DB-RMS) that is aimed at meeting such requirements. It is built on top of the UDP sub-layer. DB-RMS provides multiple service options, each with different fault detection/recovery capabilities and different costs in terms of the execution overhead. This feature makes DB-RMS suitable for a variety of applications. The service options and the corresponding support protocols are presented first, followed by a formal analysis on the detection and reaction latency bounds under various service options. A messaging layer prototype using DB-RMS has been implemented. Performance measurements of the prototype implementation have been taken and the positive results obtained are also presented.

A middleware model named ROAFTS (Real-time Object-oriented Adaptive Fault Tolerance Support) has been evolving in the UCI DREAM Lab. over the past decade as the core of a reliable execution engine model for fault-tolerant (FT) real-time (RT) distributed computing (DC) applications. It is meant to be an integration of various mechanisms for fault detection and recovery in a form that meshes well with high-level RT DC object-/component- based programming, in particular, TMO (Time-triggered Message-triggered Object) programming. Using ROAFTS as a backbone and low-layer middleware, we developed a model and a skeleton implementation for FT DC middleware providing efficient FT execution services for component-based home network applications. Capabilities for management of home information processing devices, including health monitoring of home devices, reconfiguration of device connections, and servicing queries on device status, were added to ROAFTS. Those additions were first designed as a network of high-level RT DC components, i.e., TMOs. Then the TMO network was extended into an FT TMO network by applying the replication scheme called the PSTR (Primary-Shadow TMO Replication) scheme and incorporating a component responsible for reconfiguring TMO replicas. This extension of ROAFTS is called ROAFTS-HNE (Home Network Extension) and its architecture is presented here. In addition, during the development of the ROAFTS-HNE model, we formulated a new approach for applying the PSTR scheme to RT DC components supported by ROAFTS. Finally, evaluations of the recovery times of a prototype implementation have been conducted.

Given the diversity of home environment, appliances, and residents, the applications for smart homes must be configurable and adaptive. Instead of programming each household, we propose an ontology-based framework to facilitate the automatic composition of appropriate applications. The system composes appropriate services depending upon the available equipments in each individual household automatically. Meanwhile, it dynamically adjusts the environment parameters to match the customer needs and to encompass the available resource. With its supporting on customized function template editing, customers are able to specify their usual behavior templates as different mode.

Universal middleware bridge (UMB) provides seamless interoperation among heterogeneous home network middleware. There have been high demands for the UMB components (UMB core and adaptors) to have fault- tolerance capabilities. This paper presents a TMO structuring approach together with new implementation techniques for the fault-tolerant TMO-replica structuring scheme called PSTR. PSTR implementations of UMB components provide fault tolerance capabilities essential in realizing high reliability for the UMB facility.

For the future premium home network services, the high volume of network bandwidth and QoS support is essential in the home networks. Even though ultra-wideband (UWB) is considered as a promising technology because of its high bandwidth up to 480 Mbps, it has a limitation of short transmission range. To overcome UWB short coverage, we converge here UWB and high-speed powerline communication. Here we introduce a high-speed powerline communication bridge, more specifically a HomePlug AV and WiMedia UWB bridge. We explain the functions of hardware and software structure of it. Our bridge provides not only the connectivity between two networks but also supports various types of QoS for multi-media services.

This paper proposes the design of the autonomous fault processing mechanism that can be used to solve abnormal faults in home network. In home network environments that consists of several kind of networks and devices, their unrevealed functionalities can cause non-instinct faults (combination or inferred situation). Usually many researches had tried to make a model-based fault processing mechanism. But, those works depended on the characteristics of the model for the specific environment. We focus on the process in which faults are induced and establish fault categories that can be caused in home network and the autonomous fault processing mechanism.

The QoS support is very important for the future home network services. Since home network is mixed with various types of networks, it is required to support QoS even when traffics flow over non-QoS network like Ethernet. We design and implement a bridge system providing QoS which connects non-QoS and QoS networks, more specifically Ethernet and high-speed power line communication (H-PLC). This paper explains the issues for supporting QoS and introduces a QoS bridge system for Ethernet and H-PLC.

This paper describes how the conversion of a device and a control message is accomplished in a UMB. The UMB solves interoperability problems caused by the heterogeneity of several types of home network middleware. To solve the problems, conversion of a device and messages among the middleware is indispensable. Devices can see and control other devices connected to different middlewares in the UMB framework by the mechanism of device and message conversion.

IEEE 802.11i standard supports a secure access control for wireless LAN and IEEE 802.1X standard includes various authentication method protocols. It is expected that next-generation wireless LAN security techniques will be based on IEEE 802.1X and IEEE 802.11i standards. However, at present users who are not familiar with a computer and an authentication details have difficulty to setup the network security based on IEEE 802.11i. Accordingly, this paper proposes authentication scenarios to minimize the process by users, a password method which is changed randomly and periodically, and authentication protocols. The proposed protocols provide convenience for nonprofessional computer users as well as secure home network environment against the unwanted attacks such as a brute force attack or a replay attack

This work presents the UPnP - IEEE1394 software bridge structure for representing legacy IEEE1394 devices to the UPnP middleware. This bridge implementation provides a UPnP based control mechanism for IEEE1394 devices and data transferring from IEEE1394 bus to IP network with formal conversion.

We propose the electrical crosstalk reduction method of the gigabit transceiver module for the FTTH (Fiber-To-The-Home) applications. For the electrical crosstalk reduction between LD and PD, we insert a dummy ground line nearby a signal lines.