A5009759489- Robot Manipulation and Learning
- Robotics and Sensor-Based Localization
- Robotic Path Planning Algorithms
- Social Robot Interaction and HRI
- Face recognition and analysis
- Human Motion and Animation
- Robotics and Automated Systems
- Reinforcement Learning in Robotics
- 3D Shape Modeling and Analysis
- Human Pose and Action Recognition
- Gaze Tracking and Assistive Technology
- Microgrid Control and Optimization
- Software Engineering Research
- Advanced Vision and Imaging
- Advanced Neural Network Applications
- Multimodal Machine Learning Applications
- Energy Load and Power Forecasting
- Soft Robotics and Applications
- Generative Adversarial Networks and Image Synthesis
- Prosthetics and Rehabilitation Robotics
- Autonomous Vehicle Technology and Safety
- Mental Health Treatment and Access
- Smart Grid Energy Management
- Advanced Graph Neural Networks
- Healthcare Decision-Making and Restraints
National Institute of Mental Health and Neurosciences
2024-2025
Kinetic BEI (United States)
2024
Koneru Lakshmaiah Education Foundation
2024
Nara Institute of Science and Technology
2013-2021
The University of Tokyo
2018
Bunkyo University
2018
Kyushu Institute of Technology
2017
Now a days, Generative Adversarial Networks (GANs) are an arising technology for both supervised and unsupervised learning which have capability to generate data of high standard. Image translation is one the application GANs as augmentation we used in this proposed framework. makes mapping between source image target easier it calculates loss function also improve quality generated image. In paper, Conditional translates images based upon some conditions. The performance analyzed model by...
Robotic solutions to clothing assistance can significantly improve quality of life for the elderly and disabled. Real-time estimation human-cloth relationship is crucial efficient learning motor skills robotic assistance. The major challenge involved cloth-state due inherent nonrigidity occlusion. In this study, we present a novel framework real-time cloth state using low-cost depth sensor, making it suitable feasible social implementation. relies on hypothesis that articles are constrained...
The recent demographic trend across developed nations shows a dramatic increase in the aging population, fallen fertility rates and shortage of caregivers. Hence, demand for service robots to assist with dressing which is an essential Activity Daily Living (ADL) increasing rapidly. Robotic Clothing Assistance challenging task since robot has deal two demanding tasks simultaneously, (a) non-rigid highly flexible cloth manipulation (b) safe human–robot interaction while assisting humans whose...
As the retail industry keeps expanding and shortage of workers increasing, there is a need for autonomous manipulation products to support operations. The increasing amount customers in establishments such as convenience stores requires automation restocking, disposing straightening products. needs be time-efficient, avoid damaging beautify display In this paper, we propose robotic system restock shelves, dispose expired products, straighten environments. proposed mobile manipulator features...
Real-time estimation of human-cloth relationship is crucial for efficient learning motor skills in robotic clothing assistance. However, cloth state using a depth sensor challenging problem with inherent ambiguity. To address this problem, we propose the offline dynamics model by incorporating reliable motion capture data and applying online tracking sensor. In study, evaluate performance shared Gaussian Process Latent Variable Model articles. The experimental results demonstrate...
Recent demographic trends in super aging societies, such as Japan, is leading to severe worker shortage. Service robots can play a promising role augment human workers for performing various household and assistive tasks. Toilet cleanup one challenging task that involves complaint motion planning constrained toilet setting. In this study, we propose an end-to-end robotic framework perform tasks related cleanup. Our key contributions include the design of multipurpose end-effector, adaptive...
In this study, we propose a novel method for the real-time estimation of Human-Cloth relationship, which is crucial efficient motor skill learning in Robotic Clothing Assistance. This system relies on use low cost depth sensor, provides color and images without requiring an elaborate setup making it suitable real-world applications. We present algorithm to estimate parameters that represent topological relationship between human clothing article. At core our approach are dimensional...
Learning from Demonstration (LfD) is a paradigm where humans demonstrate the procedure to perform complex tasks which can be used train autonomous agents. However, performance of LfD highly sensitive quality demonstrations in turn depends on user-interface. In this paper, we propose use Virtual Reality (VR) develop an intuitive interface that enables users provide good demonstrations. We apply approach task training visual attention system crucial component for such as driving and...
Motor-skill learning for complex robotic tasks is a challenging problem due to the high task variability. Robotic clothing assistance one such that can greatly improve quality-of-life elderly and disabled. In this study, we propose data-efficient representation encode task-specific motor-skills of robot using Bayesian nonparametric latent variable models. The effectivity proposed motor-skill demonstrated in two ways: (1) through real-time controller be used as tool from demonstration impart...
In this paper, we address the problem of solving rearranging tasks using a robot. Rearranging are challenging because they include many problems to solve at same time, such as determining how pick items well planning and where place them. Solving task usually consists finding set pick-and-place instructions with symbolic planner perform task. However, if does not consider robot's capability execute instructions, it will likely generate infeasible which wastes time in multiple trials...
A single shepherd dog can herd a flock of sheep to gate. Despite heuristic algorithm based on adaptive switching between collecting the when they are too dispersed and driving them once aggregated, it remains unknown how learns switching. In fact, reinforcement learning models have not succeeded so far in reproducing without explicitly making two strategies. Here, we show that an imitation model reproduce algorithm, is, from demonstrations by expert. We also confirmed does simply copy but...
Department of Psychiatry, Psychiatric Rehabilitation Services, National Institute Mental Health and Neuro Sciences, Bengaluru, Karnataka, India Address for correspondence: Dr. K. N. Nishanth, India. E-mail: [email protected]
As the robot technology is advancing, it possible to use robots for basic day-to-day chores, so that burden can be taken off from humans. To make perform such tasks, necessary them handle different types of objects. Manipulation deformable objects as cloth a challenging task because high dimensionality and large number configurations cloth. Previous studies have covered simple manipulations articles. In this paper, we are focusing on table setting requires putting sheet table. This paper...
In this research, an effort is made to address microgrid systems' operational challenges, characterized by power oscillations that eventually contribute grid instability. An integrated strategy proposed, leveraging the strengths of convolutional and Gated Recurrent Unit (GRU) layers. This approach aimed at effectively extracting temporal data from energy datasets improve precision behavior forecasts. Additionally, attention layer employed underscore significant features within time-series...
A robust, easy-to-deploy robot for service tasks in a real environment is difficult to construct. Record-and-playback (R&P) method used teach motor-skills robots performing tasks. However, R&P methods do not scale challenging where even slight changes the environment, such as localization errors, would either require trajectory modification or new demonstration. In this paper, we propose Sequence-to-Sequence (Seq2Seq) based neural network model generate trajectories configuration space given...
In this paper, we address the task of rearranging items with a robot. A is challenging because one should solve following issues: to determine how pick and plan where place items. study, focus on obtain sequence actions that robot could execute reducing failures when motion planner creates trajectory move robot, such as not finding solution. To confirm instructions before executing them combine symbolic planner. For purpose, propose Motion Feasibility Checker (MFC), which quickly decides if...
Synthetic image generation plays a crucial role in the development of robot vision algorithms, circumventing manual data collection. However, realism synthetic images could affect performance algorithms when applied real-world settings. In this study, we propose framework to quantitatively assess using set metrics as means characterization. We use commercial rendering engine test-bed for generating and ascertain that parameters specific through statistical hypothesis testing. demonstrate can...