Article
Computer Science, Software Engineering
Jorge Wagner, Wolfgang Stuerzlinger, Luciana Nedel
Summary: In this study, two standard interaction techniques for Immersive Analytics environments were evaluated: virtual hands and virtual ray pointers, as well as a mixed mode. The mixed mode was found to significantly reduce completion times for tasks in the spatio-temporal data domain, with a slight decrease in overall success rates.
IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS
(2021)
Article
Computer Science, Interdisciplinary Applications
Haiyan Jiang, Dongdong Weng, Zhen Song, Xiaonuo Dongye, Zhenliang Zhang
Summary: This paper proposes a neural network-based approach for generating finger movements that interact with objects. The experimental results show that the method can generate plausible hand manipulation motions without noticeable delay.
Article
Computer Science, Information Systems
Hyeongil Nam, Chanhee Kim, Kangsoo Kim, Jong-Il Park
Summary: In this paper, a kinematics-based realistic hand interaction method is proposed to enable a physically plausible grip-lifting experience in VR. A human subjects experiment using a prototype shows positive effects on the perceived realism and usefulness of the interaction. The method contributes to the design and development of realistic virtual experiences.
Article
Computer Science, Cybernetics
Xiaolong Liu, Lili Wang, Shuai Luan
Summary: Object manipulation is crucial in virtual reality and especially important in collaborative VR applications. This study introduces a collaborative method using the manipulation guidance field (MGF) to improve accuracy and efficiency. The MGF guides users to different viewpoints for efficient and collaborative object manipulation. The results of a user study demonstrate that the MGF-based method significantly reduces task completion time, errors, and task load compared to a control method without MGF.
INTERNATIONAL JOURNAL OF HUMAN-COMPUTER INTERACTION
(2023)
Article
Computer Science, Cybernetics
Qinwen Zheng, Lili Wang, Wei Ke, Sio Kei Im
Summary: This paper proposes a virtual reality object manipulation method based on a variable virtual interaction region. By introducing a hand interaction hemisphere region and an interaction heat volume-based method, the proposed method achieves significant improvements in task completion time, manipulation precision, and user comfort. It also outperforms other methods in terms of task load and usability.
INTERNATIONAL JOURNAL OF HUMAN-COMPUTER INTERACTION
(2023)
Article
Engineering, Mechanical
Rajesh Kumar, Sudipto Mukherjee
Summary: This article presents a method for relocating robotic fingertips on the surface of an object during precision grasp. It proposes a metric to limit deviation from force closure during repositioning and describes a manipulability-based metric to search for the optimal goal grasp states. The manipulability-based metric is used to increase the range of object motion by relocating the contacts.
JOURNAL OF MECHANISMS AND ROBOTICS-TRANSACTIONS OF THE ASME
(2022)
Article
Chemistry, Analytical
Patricio Rivera, Edwin Valarezo Anazco, Tae-Seong Kim
Summary: The study proposes a deep reinforcement learning approach to train an anthropomorphic robotic hand for natural grasping and relocation tasks using a synergy space. The synergy-based DRL achieves a higher success rate in object manipulation tasks compared to standard DRL methods.
Article
Automation & Control Systems
Guotao Li, Xu Liang, Yifan Gao, Tingting Su, Zhijie Liu, Zeng-Guang Hou
Summary: A linkage-driven underactuated three-finger robotic hand is proposed in this paper to imitate the f/e and a/a motions of the human hand. The robotic hand consists of three identical underactuated fingers with a spherical four-bar mechanism and bevel gears. The kinematic model of the spherical mechanism is established based on screw theory, and the maximum available workspace index (MAW) is proposed to evaluate the workspace. The parameters of the spherical mechanism are optimized to improve the performance of the robotic hand, and experiments show its capability for adaptive grasping and in-hand manipulation.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2023)
Article
Robotics
Michael A. Lin, Rachel Thomasson, Gabriela Uribe, Hojung Choi, Mark Cutkosky
Summary: The study introduces a new gripper and exploration approach that utilizes a finger with light touch for probing and grasping objects. Experimental results demonstrate that the finger can safely and quickly make contact with objects, leveraging contact information to reduce uncertainty in the environment.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Computer Science, Artificial Intelligence
Juntong Yun, Gongfa Li, Du Jiang, Manman Xu, Feng Xiang, Li Huang, Guozhang Jiang, Xin Liu, Yuanmin Xie, Bo Tao, Zifan Fang
Summary: This study proposes a digital twin model construction method for robot grasping based on digital twin technology, aiming to solve the problem of object grasping in a multi-object stacking environment. By establishing the logical relationship between the robot digital twin model and the virtual model of stacked objects, grasping planning and virtual-real interaction are achieved.
APPLIED SOFT COMPUTING
(2023)
Article
Robotics
Li Tian, Hanhui Li, Qifa Wang, Xuezeng Du, Jialin Tao, Jordan Sia Chong, Nadia Magnenat Thalmann, Jianmin Zheng
Summary: The study introduces a robotic hand design framework based on human gestures, utilizing 33 grasping gestures as the basis for design and implementation, and using frame interpolation to accomplish manipulation tasks. Experimental results validate the dexterity of the design.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Robotics
Haizhuang Jiang, Xin Jiang, Yanling Zhou, Haoyao Chen, Peng Li, Yunhui Liu
Summary: This study proposes an in-hand manipulation method for fabrics based on variable friction and verifies its effectiveness through experiments.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Computer Science, Interdisciplinary Applications
Sebastian Marichal, Inigo Ezcurdia, Rafael Morales, Amalia Ortiz, Asier Marzo, Oscar Ardaiz
Summary: In virtual reality environments, haptic feedback is almost as important as visual information. Active haptic feedback requires expensive specialized devices, while passive haptic feedback uses inexpensive objects as substitutes for virtual entities. This study proposes the Hand-as-a-Prop technique, using human hands as object props, which can represent multiple shapes without extra hardware.
Article
Robotics
Jing Huang, K. W. Samuel Au
Summary: The grasping mechanisms in robotic deformable object manipulation are not well-studied due to the complex nature of this field. This study focuses on the problem of grasping position selection in deformable object manipulation, which is crucial for the feasibility and efficiency of the tasks. Task-oriented methods have been developed for selecting appropriate grasping positions, taking into account the properties of both the task and deformable object. Evaluation of these methods has been performed through simulations and hardware experiments.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Computer Science, Software Engineering
Giorgos Ganias, Christos Lougiakis, Akrivi Katifori, Maria Roussou, Yannis Ioannidis, Ioannis Panagiotis Ioannidis
Summary: Virtual grasping is crucial in Virtual Environments, but there are few studies focusing on handheld controllers. In this study, three different grasping visualizations were compared in immersive VR using controllers. The results showed that the Auto-Pose (AP) visualization was preferred by users and enhanced the perceived sense of embodiment. This study encourages the inclusion of similar visualizations in future research and VR experiences.
IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS
(2023)
Article
Automation & Control Systems
Bharath Gopalakrishnan, Arun Kumar Singh, K. Madhava Krishna, Dinesh Manocha
Summary: This article presents an efficient algorithm for solving chance-constrained optimization under nonparametric uncertainty by representing distributions as functions in Reproducing Kernel Hilbert Space (RKHS) and minimizing the distance between desired and constraint functions. The approach constructs the desired distribution based on scenario approximation and uses the kernel trick to simplify computational complexity. Validation on robotic applications with nonlinear and nonconvex chance constraints shows significant improvements in sample complexity and achieved optimal cost compared to existing approaches.
IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY
(2022)
Article
Engineering, Civil
Rohan Chandra, Aniket Bera, Dinesh Manocha
Summary: Research shows that autonomous vehicles can be socially aware if there is a mechanism to understand human driver behavior. A new approach using machine learning to predict human driver behavior is presented, which extracts driver behavior features and creates a computational mapping between vehicle trajectories and driver behaviors. The method is proven to be robust, general, and applicable to various autonomous navigation scenarios, with evaluations conducted on real-world traffic datasets and simulations.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Chaochao Li, Pei Lv, Dinesh Manocha, Hua Wang, Yafei Li, Bing Zhou, Mingliang Xu
Summary: This article investigates antagonistic crowd behaviors, analyzes the role of antagonistic emotions and emotional contagion in crowd violence, and combines evolutionary game theory with it. The research shows that the model can predict crowd behaviors and has been validated in real-world scenarios.
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING
(2022)
Article
Robotics
Adarsh Jagan Sathyamoorthy, Utsav Patel, Moumita Paul, Nithish K. Sanjeev Kumar, Yash Savle, Dinesh Manocha
Summary: This paper presents a novel approach called CoMet for computing a group's cohesion and using it to improve a robot's navigation in crowded scenes. The authors propose a cohesion-metric that builds on prior work in social psychology and compute this metric by utilizing visual features of pedestrians. They design and improve a navigation scheme based on this cohesion-metric and evaluate its performance on various metrics, showing significant decreases in freezing rate, deviation, and path length of the trajectory.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Robotics
Rohan Chandra, Dinesh Manocha
Summary: This article presents a new method for multi-agent planning involving human drivers and autonomous vehicles in unsignaled intersections, roundabouts, and during merging. The method utilizes game theory to develop an auction-based approach that determines the optimal action for each agent based on their driving style, effectively preventing collisions and deadlocks.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Computer Science, Software Engineering
Jin Dai, Xinyu Zhang
Summary: We propose an improved image caption generation model that incorporates a multimodal attention mechanism, channel attention mechanism, and spatial attention mechanism to enhance the model's ability to learn and utilize the grammatical features of natural sentences. We accelerate the model training using GPU parallel computing and apply the model to early education scenarios, specifically show and tell for kids. Experimental results show that our model improves captioning accuracy in terms of standard automatic evaluation metrics.
COMPUTER ANIMATION AND VIRTUAL WORLDS
(2022)
Article
Robotics
Tianrui Guan, Divya Kothandaraman, Rohan Chandra, Adarsh Jagan Sathyamoorthy, Kasun Weerakoon, Dinesh Manocha
Summary: We propose GA-Nav, a novel group-wise attention mechanism that can identify safe and navigable regions in unstructured environments from RGB images. Our method extracts multi-scale features from each type of terrain independently and classifies terrains based on their navigability levels. Integrating GA-Nav with deep reinforcement learning-based navigation algorithm improves navigation success rate and trajectory selection in real-world unstructured terrains.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Robotics
Qiaoyun Wu, Jun Wang, Jing Liang, Xiaoxi Gong, Dinesh Manocha
Summary: The method presents a new approach for image-goal navigation, decoupling the learning of navigation goal planning, collision avoidance, and navigation ending prediction. By using four separate modules, the method improves navigation success rate and reduces collision rate in real complex environments compared to some state-of-the-art models.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Robotics
Jing Liang, Kasun Weerakoon, Tianrui Guan, Nare Karapetyan, Dinesh Manocha
Summary: We propose a novel outdoor navigation algorithm that generates stable and efficient robot actions to reach a goal. Our approach, based on the Proximal Policy Optimization (PPO) algorithm, achieves multiple capabilities for outdoor local navigation tasks, such as reducing drifting, maintaining stability on bumpy terrains, avoiding steep hills, and preventing collisions. By training with rich features from a Lidar sensor in a high-fidelity Unity simulator, our method mitigates the gap between simulation and the real world. Evaluation results demonstrate significant improvements in stability, drifting reduction, and elevation changes compared to state-of-the-art approaches.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Robotics
Hao Tian, Chaoyang Song, Changbo Wang, Xinyu Zhang, Jia Pan
Summary: We propose an incremental sampling-based task and motion planner for retrieving near-cylindrical objects in cluttered scenes. Our method computes collision-free motion for a robot to retrieve the target object by removing obstacles in a planned sequence. Experimental results demonstrate the efficiency and effectiveness of our method in solving high-dimensional planning problems.
IEEE TRANSACTIONS ON ROBOTICS
(2023)
Article
Robotics
Vrushabh Zinage, Senthil Hariharan Arul, Dinesh Manocha, Satadal Ghosh
Summary: We propose an online motion planning algorithm (3DOGSSE) that generates smooth and collision-free trajectories for a 3-D agent operating in an unknown, obstacle-cluttered environment. The algorithm constructs an obstacle-free region called generalized sensed shape in each planning iteration and computes a collision-free path within it. The generated trajectory is constrained to lie within the generalized sensed shape, ensuring obstacle-free maneuvering for the agent.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Computer Science, Software Engineering
James F. F. Mullen Jr, Dinesh Manocha
Summary: In this study, a novel method called PACE is presented for modifying motion-captured virtual agents to interact with and move throughout dense, cluttered 3D scenes. The method optimizes the motion of the virtual agent by adjusting it according to the obstacles and objects in the environment. The authors compared their method with prior motion generating techniques and found that their method outperformed the others in terms of human preference and physical plausibility metrics.
IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Divya Kothandaraman, Tianrui Guan, Xijun Wang, Shuowen Hu, Ming Lin, Dinesh Manocha
Summary: This paper presents Fourier Activity Recognition (FAR), an algorithm for UAV video activity recognition. The algorithm utilizes a novel Fourier object disentanglement method to separate the human agent from the background. It also introduces Fourier Attention algorithm to capture contextual information and long-range space-time dependencies. Experimental results demonstrate significant improvements in accuracy and speed over prior works.
COMPUTER VISION, ECCV 2022, PT XXXVII
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Jiangbei Yue, Dinesh Manocha, He Wang
Summary: In this paper, a new method combining model-based and model-free techniques for trajectory prediction is proposed. By using an explicit physics model and a deep neural network, the method achieves excellent performance in modeling pedestrian behaviors and data fitting. The method also demonstrates better generalizability in different scenarios and provides explanations for pedestrian behaviors.
COMPUTER VISION, ECCV 2022, PT XXXIV
(2022)