Article
Robotics
Fabian Reister, Markus Grotz, Tamim Asfour
Summary: Mobile manipulation tasks require seamless integration of navigation and manipulation capabilities. This study proposes a method that autonomously selects optimal robot placements and grasp candidates for efficient task execution, considering both navigation costs and manipulation costs. The method is evaluated in simulated and real-world experiments, showcasing its effectiveness and the ability to update optimal placements in dynamic environments.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Robotics
Eric Pairet, Constantinos Chamzas, Yvan Petillot, Lydia E. Kavraki
Summary: This paper introduces a new planning method that effectively utilizes previous experiences in diverse task instances. Compared to traditional experience-based planners, our planners outperform them in both success rate and planning time by decomposing and reshaping prior experiences to compose valid motion plans.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Computer Science, Artificial Intelligence
Xiaotong Hua, Guolei Wang, Jing Xu, Ken Chen
Summary: This paper presents a path planner based on reinforcement learning skills, which achieves robust planning for duct-enter tasks by incorporating optimization functions, robot end orientation, and path guide points into the action part. Experimental results show a significant improvement in success rate compared to traditional methods.
JOURNAL OF INTELLIGENT MANUFACTURING
(2021)
Article
Robotics
Jixuan Zhi, Lap-Fai Yu, Jyh-Ming Lien
Summary: The environment plays a significant role in human-robot interactions, impacting safety and comfort for humans and effectiveness and efficiency for robots. Redesigning spaces can enhance collaboration and interactions between humans and robots.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Robotics
Benjamin Riviere, Wolfgang Honig, Matthew Anderson, Soon-Jo Chung
Summary: This study introduces a Neural Tree Expansion (NTE) method for multi-robot online planning, which adapts the AlphaZero method to a decentralized, partial information, continuous action space setting in order to achieve coordination between robots and demonstrate computational advantages in experiments.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Robotics
Teguh Santoso Lembono, Emmanuel Pignat, Julius Jankowski, Sylvain Calinon
Summary: The study introduces a method using generative adversarial network to learn the distribution of valid robot configurations under constraints, applied in inverse kinematics and motion planning with simulation validation.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Robotics
Nikhil Das, Michael C. Yip
Summary: This study proposes the use of Gaussian process regression and forward kinematics kernel to efficiently and accurately estimate collision distance for robot manipulators. The GP model with FK kernel achieves significantly faster distance evaluations compared to standard geometric techniques and more accurate evaluations than other regression models. A confidence-based hybrid model is also introduced, showing usefulness in different areas by switching between model-based predictions and sensor-based approaches.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Robotics
Hai Zhu, Francisco Martinez Claramunt, Bruno Brito, Javier Alonso-Mora
Summary: This letter introduces a data-driven decentralized trajectory optimization approach for multi-robot motion planning in dynamic environments. By incorporating a novel trajectory prediction model based on recurrent neural networks (RNN) into a decentralized model predictive control (MPC) framework, the approach is able to achieve predictive collision avoidance without relying on communication, showing comparable performance to a centralized planner and scalability to a large number of robots.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Robotics
Nils Funk, Juan Tarrio, Sotiris Papatheodorou, Marija Popovic, Pablo F. Alcantarilla, Stefan Leutenegger
Summary: The proposed system utilizes adaptive-resolution volumetric mapping concept integrating with the hierarchical decomposition in an octree data structure, enabling fast collision queries for robot motion planning, and showing improvements in mapping accuracy and other aspects over existing techniques, particularly for high-resolution maps.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Chemistry, Multidisciplinary
Adrian Prados, Alicia Mora, Blanca Lopez, Javier Munoz, Santiago Garrido, Ramon Barber
Summary: The advancement of robotics has led to the growth of complex robotic applications. This paper presents a kinesthetic learning method based on fast marching square that ensures convergence and learns from the experience of a human demonstrator. An auto-learning functionality is introduced, allowing the algorithm to explore unknown states of the environment. The proposed method is evaluated through simulations and tests in different scenarios, proving its effectiveness.
APPLIED SCIENCES-BASEL
(2023)
Article
Robotics
Zhuoqi Zheng, Chao Cao, Jia Pan
Summary: This paper proposes a hierarchical approach for effective exploration and collision-free navigation of mobile robots in crowded environments. By combining local and global information and utilizing a reinforcement learning-based obstacle avoidance algorithm, the proposed method enables safe exploration in pedestrian crowds.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Computer Science, Artificial Intelligence
Steve Macenski, Shrijit Singh, Francisco Martin, Jonatan Gines
Summary: The accelerated deployment of service robots has led to the development of various algorithm variations to better handle real-world conditions. Many local trajectory planning techniques have been successfully implemented in practical robot systems. However, the use of pure path tracking algorithms is still prevalent, with few variants considering variable linear velocities. This paper introduces a regulated variant of the Pure Pursuit algorithm that incorporates additional heuristics to adjust linear velocities, with a focus on safety in constrained and partially observable spaces commonly encountered by deployed robots.
Article
Robotics
Theodoros Stouraitis, Michael Gienger
Summary: This letter presents a novel concept to support physically impaired humans in daily object manipulation tasks with a robot, by proposing a predictive model and encoding constancy constraints for considering dependencies between sequential behavior. Numerical studies, simulations and robot experiments were conducted to analyse and evaluate the proposed method in table top tasks.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Robotics
Yixin Lin, Austin S. Wang, Eric Undersander, Akshara Rai
Summary: In this research, manipulation tasks are approached as a graph classification problem, and a graph neural network (GNN) policy architecture is employed to solve these tasks based on demonstrations. Experimental results demonstrate that a GNN policy trained through imitation learning can successfully solve blockstacking, rearrangement, and dishwasher loading tasks, and can generalize to a wider range of scenarios.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Robotics
Parikshit Maini, Burak M. Gonultas, Volkan Isler
Summary: This paper presents the design of Cowbot, an autonomous weed mowing robot for maintaining cow pastures. The robot utilizes online planning to detect and mow weeds, optimizing path length based on real-time information. Field experiments show a 60% reduction in path length compared to traditional methods.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Immunology
Renata Fioravanti Tarabini, Mauricio Menegatti Rigo, Andre Faustino Fonseca, Felipe Rubin, Rafael Belle, Lydia E. Kavraki, Tiago Coelho Ferreto, Dinler Amaral Antunes, Ana Paula Duarte de Souza
Summary: The COVID-19 pandemic caused by the SARS-CoV-2 coronavirus has resulted in over 4.5 million deaths worldwide. Despite the development of vaccines, the virus continues to spread globally due to resistance to vaccination and limited access to vaccines. Developing additional therapeutic strategies is crucial to combat SARS-CoV-2 and other coronaviruses.
FRONTIERS IN IMMUNOLOGY
(2022)
Editorial Material
Cell & Tissue Engineering
Jason L. Guo, Lydia E. Kavraki, Antonios G. Mikos
TISSUE ENGINEERING PART A
(2023)
Article
Robotics
Zachary Kingston, Lydia E. Kavraki
Summary: Robotic manipulation often involves multimodal planning tasks, which require finding a sequence of transitions between modes. However, many multimodal planners fail to scale when faced with difficult motion planning or tasks with a long horizon. This work proposes a solution that uses experience-based planning and a layered planning approach to improve the scalability and task satisfaction of multimodal planners, enabling them to handle complex manipulation tasks and achieve significant improvements in high-dimensional robot scenes.
IEEE TRANSACTIONS ON ROBOTICS
(2023)
Article
Robotics
Valentin N. Hartmann, Andreas Orthey, Danny Driess, Ozgur S. Oguz, Marc Toussaint
Summary: Robotic construction assembly planning is a parallelizable task and motion planning problem. We propose a planning system that parallelizes complex task and motion planning by solving smaller subproblems. By combining optimization methods and a sampling-based path planner, we can plan cooperative multi-robot manipulation with unknown arrival times. We demonstrate the robustness and scalability of this approach in multiple construction case studies and showcase the feasibility of executing the computed plans in the real world.
IEEE TRANSACTIONS ON ROBOTICS
(2023)
Article
Robotics
Christos K. Verginis, Dimos Dimarogonas, Lydia E. Kavraki
Summary: We introduce KDF, a new framework for solving the kinodynamic motion-planning problem via funnel control by integrating sampling-based planning techniques with funnel-based feedback control. The proposed scheme is easily distributable to various systems and scenarios without requiring knowledge of the system's dynamics. The safety of the system is ensured through a high-level safe path and a low-level funnel control algorithm.
IEEE TRANSACTIONS ON ROBOTICS
(2023)
Article
Immunology
Sarah Hall-Swan, Jared Slone, Mauricio M. Rigo, Dinler A. Antunes, Gregory Lizee, Lydia E. Kavraki
Summary: PepSim is a method for predicting T-cell cross-reactivity based on the structural and biochemical similarity of pHLAs. It accurately separates cross-reactive from non-crossreactive pHLAs in diverse datasets, making it a valuable tool for designing safe and effective T-cell immunotherapies.
FRONTIERS IN IMMUNOLOGY
(2023)
Article
Chemistry, Multidisciplinary
Eleni E. Litsa, Vijil Chenthamarakshan, Payel Das, Lydia E. Kavraki
Summary: The authors present a deep learning architecture for recommending molecular structures based on mass spectra alone. This method has applications in chemical compound identification and drug discovery.
COMMUNICATIONS CHEMISTRY
(2023)
Article
Biochemical Research Methods
Anja Conev, Mauricio Menegatti Rigo, Didier Devaurs, Andre Faustino Fonseca, Hussain Kalavadwala, Martiela Vaz de Freitas, Cecilia Clementi, Geancarlo Zanatta, Dinler Amaral Antunes, Lydia E. Kavraki
Summary: Proteins are dynamic macromolecules that play vital roles in cells, and understanding their conformational landscapes is crucial for understanding their function. Our new approach, EnGens, provides a unified framework for generating and analyzing representative protein conformational ensembles from available structural datasets. These representative ensembles can be used for various downstream tasks such as protein-ligand docking, protein dynamics modeling, and analysis of mutations.
BRIEFINGS IN BIOINFORMATICS
(2023)
Article
Computer Science, Software Engineering
Michael Gleicher, Maria Riveiro, Tatiana von Landesberger, Oliver Deussen, Remco Chang, Christina Gillman, Theresa-Marie Rhyne
Summary: Visualization researchers and professionals seek appropriate abstractions to consider visualization solutions independently from specific problems. This article introduces a problem space that complements existing frameworks, focusing on the needs a visualization is meant to address. It provides a valuable conceptual tool for designing and discussing visualizations.
IEEE COMPUTER GRAPHICS AND APPLICATIONS
(2023)
Article
Robotics
Servet B. Bayraktar, Andreas Orthey, Zachary Kingston, Marc Toussaint, Lydia E. Kavraki
Summary: Rearrangement puzzles are variants of rearrangement problems that involve logically linked elements. To efficiently solve such puzzles, a motion planning approach based on a logically factored state space is developed, integrating the robot's capabilities through factors of simultaneously manipulatable joints. A planner called LA-RRT is proposed, which optimizes for a low number of actions. A new path defragmentation method is at the core of the approach, minimizing action cost by rearranging and optimizing consecutive edges. The performance of LA-RRT is significantly better than the next best asymptotically-optimal planner, achieving 4.01 to 6.58 times improvement in final action cost.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Proceedings Paper
Robotics
Khen Elimelech, Lydia E. Kavraki, Moshe Y. Vardi
Summary: This paper explains how to use a library of abstract skills derived from past planning experience to reduce the computational cost of solving new task planning problems. It shows how matching skills to tasks allows for task decomposition and parallel solving, and provides a hierarchical solution algorithm that integrates with any standard task planner.
ROBOTICS RESEARCH, ISRR 2022
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Pragathi Praveena, Michael Gleicher, Bilge Mutlu
Summary: This paper explores the potential of collaborative robots as intelligent embodied agents for remote human collaboration. The authors discuss their iterative design process to develop interaction techniques that distribute camera control between the robot and human collaborators using shared control-based methods.
COMPANION OF THE ACM/IEEE INTERNATIONAL CONFERENCE ON HUMAN-ROBOT INTERACTION, HRI 2023
(2023)
Proceedings Paper
Mathematics, Applied
Khen Elimelech, Lydia E. Kavraki, Moshe Y. Vardi
Summary: Solving realistic robotic task planning problems is computationally demanding. To increase the reusability of successful plans and reduce future planning cost, a systematic and automatable approach for plan transfer is proposed. This approach suggests caching successful plans in a dynamically-defined abstract domain as abstract skills, which allows for a unified, standardized, and compact skill database and lifelong operation.
ALGORITHMIC FOUNDATIONS OF ROBOTICS XV
(2023)
