4.7 Article

A Deep Reinforcement Learning Approach to Optimal Morphologies Generation in Reconfigurable Tiling Robots

Related references

Note: Only part of the references are listed.
Article Computer Science, Interdisciplinary Applications

Multi-fidelity reinforcement learning framework for shape optimization

Sahil Bhola et al.

Summary: We address the limitation of conventional deep reinforcement learning methods by introducing a controlled transfer learning framework that leverages a multi-fidelity simulation setting. Our strategy is applied to an airfoil shape optimization problem at high Reynolds numbers, where it learns an optimal policy for generating efficient airfoil shapes by gathering knowledge from multi-fidelity environments and reduces computational costs by over 30%. Furthermore, our formulation promotes policy exploration and generalization to new environments, preventing over-fitting to data from solely one fidelity. Our results demonstrate the applicability of this framework to other scientific deep reinforcement learning scenarios where multi-fidelity environments can be used for policy learning.

JOURNAL OF COMPUTATIONAL PHYSICS (2023)

Article Multidisciplinary Sciences

A reinforcement learning approach to airfoil shape optimization

Thomas P. Dussauge et al.

Summary: Airfoil shape optimization is a challenging task due to the complexity of fluid mechanics and the high-dimensional design space. Current approaches are data-inefficient and computationally expensive. This study proposes a deep reinforcement learning approach to airfoil shape optimization, demonstrating promising results and the relevance of DRL in this field.

SCIENTIFIC REPORTS (2023)

Article Mathematics

Optimal Morphologies of n-Omino-Based Reconfigurable Robot for Area Coverage Task Using Metaheuristic Optimization

Manivannan Kalimuthu et al.

Summary: Reconfigurable robots based on polyominos or n-Omino are increasingly being explored in cleaning and maintenance tasks. This paper proposes a new strategy to select the optimal morphologies of n-Omino-based robots that maximize area coverage and minimize energy consumption.

MATHEMATICS (2023)

Article Computer Science, Artificial Intelligence

A metaheuristic approach to optimal morphology in reconfigurable tiling robots

Manivannan Kalimuthu et al.

Summary: This paper presents a metaheuristic-based framework to identify the optimal morphology of a reconfigurable robot in order to maximize area coverage and minimize energy consumption. The proposed approach uses three different metaheuristic algorithms, SMPSO, NSGA-II, and MACO, to generate the optimal morphology for every unique layout of a two-dimensional grid map by considering the path-length as the energy consumption. The experimental results demonstrate that the framework efficiently produces energy-efficient morphologies.

COMPLEX & INTELLIGENT SYSTEMS (2023)

Article Robotics

Energy-Efficient Path Planning of Reconfigurable Robots in Complex Environments

Phone Thiha Kyaw et al.

Summary: This article proposes a novel algorithm for planning energy-efficient and collision-free paths for reconfigurable robots. The algorithm combines BIT*, an informed planner, with energy-based objectives that consider the energy cost of each reconfigurable action. Additionally, it improves the convergence rate of the algorithm by enhancing the direct sampling technique of informed RRT*. Experimental results on a tetromino hinged-based reconfigurable robot demonstrate the effectiveness of the proposed path planning technique.

IEEE TRANSACTIONS ON ROBOTICS (2022)

Article Engineering, Mechanical

Reconfigurable Robotic System Design With Application to Cleaning and Maintenance

A. A. Hayat et al.

Summary: This paper proposes a heuristic framework for designing reconfigurable robots, applying established transformation principles to system and subsystem design for innovation. Experiments demonstrate that this design method enables high performance in multiple modalities.

JOURNAL OF MECHANICAL DESIGN (2022)

Article Robotics

S-Sacrr: A Staircase and Slope Accessing Reconfigurable Cleaning Robot and its Validation

Prabakaran Veerajagadheswar et al.

Summary: This letter introduces a novel robotic platform called s-Sacrr, which can change its morphology to traverse different surfaces for cleaning tasks on staircases and slopes. The robot has a reconfiguring feature that allows it to perform area coverage on various structures. Experimental results demonstrate the robot's ability to smoothly traverse and cover staircases and inclined surfaces.

IEEE ROBOTICS AND AUTOMATION LETTERS (2022)

Article Nanoscience & Nanotechnology

Single-step deep reinforcement learning for two- and three-dimensional optimal shape design

H. Ghraieb et al.

Summary: This research assesses the capabilities of deep reinforcement learning in direct optimal shape design in computational fluid dynamics systems. By using policy-based optimization and an in-house finite element environment, the study successfully reduces drag and increases lift-to-drag ratio for equivalent cylinder and ellipse shapes, respectively.

AIP ADVANCES (2022)

Article Robotics

Energy-Based Legged Robots Terrain Traversability Modeling via Deep Inverse Reinforcement Learning

Lu Gan et al.

Summary: This work presents a deep inverse reinforcement learning method for legged robots terrain traversability modeling using both exteroceptive and proprioceptive sensory data. By incorporating robot-specific inertial features, the model fidelity is improved and a reward dependent on the robot's state during deployment is provided. The proposed method utilizes the Maximum Entropy Deep Inverse Reinforcement Learning algorithm and trajectory ranking loss to optimize legged robot demonstrations. The evaluation is conducted using a dataset from an MIT Mini-Cheetah robot and a Mini-Cheetah simulator.

IEEE ROBOTICS AND AUTOMATION LETTERS (2022)

Article Computer Science, Interdisciplinary Applications

Direct shape optimization through deep reinforcement learning

Jonathan Viquerat et al.

Summary: Deep Reinforcement Learning (DRL) has achieved remarkable achievements in various domains within physics and engineering, but there is still much to be explored before the capabilities of these methods are well understood. This paper presents the first application of DRL to direct shape optimization, demonstrating that an artificial neural network trained through DRL can generate optimal shapes autonomously, paving the way to new generic shape optimization strategies in fluid mechanics and other domains where relevant reward functions can be defined.

JOURNAL OF COMPUTATIONAL PHYSICS (2021)

Article Computer Science, Artificial Intelligence

Modular deep reinforcement learning from reward and punishment for robot navigation

Jiexin Wang et al.

Summary: Modular Reinforcement Learning decomposes tasks into sub-goals for parallel learning, showing improvements in safety and efficiency compared to conventional Q-learning. New signal scaling and automatic weighting schemes have been proposed for maze-solving navigation tasks, leading to effective results in both simulation and real robot experiments.

NEURAL NETWORKS (2021)

Article Computer Science, Interdisciplinary Applications

A knowledge-enhanced deep reinforcement learning-based shape optimizer for aerodynamic mitigation of wind-sensitive structures

Shaopeng Li et al.

Summary: This study employs reinforcement learning with deep neural network for shape optimization of wind-sensitive structures and leverages domain knowledge to enhance training efficiency. The proposed optimization method outperforms existing algorithms in mitigating the effects of wind on structures.

COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING (2021)

Review Computer Science, Information Systems

Deep Reinforcement Learning Based Mobile Robot Navigation: A Review

Kai Zhu et al.

Summary: This paper discusses the application of Deep Reinforcement Learning (DRL) in mobile robot navigation, compares the relationships and differences between four typical application scenarios, describes the development of DRL-based navigation, and addresses the challenges and possible solutions facing DRL-based navigation.

TSINGHUA SCIENCE AND TECHNOLOGY (2021)

Article Chemistry, Analytical

Coverage Path Planning Using Reinforcement Learning-Based TSP for hTetran-A Polyabolo-Inspired Self-Reconfigurable Tiling Robot

Anh Vu Le et al.

Summary: The paper introduces a complete path planning (CPP) for a polyabolo tiled robot hTetran based on a TSP-based reinforcement learning optimization. It trains a reinforcement learning-based travel sales problem (TSP) with proximal policy optimization (PPO) algorithm to generate optimal trajectories that enhance the robot's navigation with minimal energy and time spent. This CPP demonstrates the ability to achieve an ideal Pareto optima trajectory in comparison to conventional techniques.

SENSORS (2021)

Article Robotics

Learning dexterous in-hand manipulation

Marcin Andrychowicz et al.

INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH (2020)

Article Construction & Building Technology

Complete coverage path planning using reinforcement learning for Tetromino based cleaning and maintenance robot

Anirudh Krishna Lakshmanan et al.

AUTOMATION IN CONSTRUCTION (2020)

Article Engineering, Electrical & Electronic

Voronoi-Based Multi-Robot Autonomous Exploration in Unknown Environments via Deep Reinforcement Learning

Junyan Hu et al.

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY (2020)

Proceedings Paper Automation & Control Systems

FreeBOT: A Freeform Modular Self-reconfigurable Robot with Arbitrary Connection Point - Design and Implementation

Guanqi Liang et al.

2020 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS) (2020)

Proceedings Paper Automation & Control Systems

Robotic Grasping using Deep Reinforcement Learning

Shirin Joshi et al.

2020 IEEE 16TH INTERNATIONAL CONFERENCE ON AUTOMATION SCIENCE AND ENGINEERING (CASE) (2020)

Article Computer Science, Information Systems

Overcoming Obstacles With a Reconfigurable Robot Using Reinforcement Learning

Liran Yehezkel et al.

IEEE ACCESS (2020)

Article Computer Science, Information Systems

Reinforcement Learning-Based Energy-Aware Area Coverage for Reconfigurable hRombo Tiling Robot

Anh Vu Le et al.

IEEE ACCESS (2020)

Article Computer Science, Information Systems

Tradeoff Between Area Coverage and Energy Usage of a Self-Reconfigurable Floor Cleaning Robot Based on User Preference

M. A. Viraj J. Muthugala et al.

IEEE ACCESS (2020)

Article Computer Science, Information Systems

A Framework for Taxonomy and Evaluation of Self-Reconfigurable Robotic Systems

Ning Tan et al.

IEEE ACCESS (2020)

Article Computer Science, Information Systems

hTetro-Infi: A Reconfigurable Floor Cleaning Robot With Infinite Morphologies

S. M. Bhagya P. Samarakoon et al.

IEEE ACCESS (2020)

Article Automation & Control Systems

Solving the optimal path planning of a mobile robot using improved Q-learning

Ee Soong Low et al.

ROBOTICS AND AUTONOMOUS SYSTEMS (2019)

Article Computer Science, Interdisciplinary Applications

Framework for design optimization using deep reinforcement learning

Kazuo Yonekura et al.

STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION (2019)

Article Construction & Building Technology

Design and modelling of a modular window cleaning robot

M. Vega-Heredia et al.

AUTOMATION IN CONSTRUCTION (2019)

Article Engineering, Aerospace

Aerodynamic shape optimization using a novel optimizer based on machine learning techniques

Xinghui Yan et al.

AEROSPACE SCIENCE AND TECHNOLOGY (2019)

Article Computer Science, Information Systems

Graph Theory-Based Approach to Accomplish Complete Coverage Path Planning Tasks for Reconfigurable Robots

Ku Ping Cheng et al.

IEEE ACCESS (2019)

Proceedings Paper Automation & Control Systems

Panthera: Design of a Reconfigurable Pavement Sweeping Robot

Abdullah A. Hayat et al.

2019 INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA) (2019)

Article Engineering, Electrical & Electronic

Methods for interpreting and understanding deep neural networks

Gregoire Montavon et al.

DIGITAL SIGNAL PROCESSING (2018)

Article Construction & Building Technology

Glass facade cleaning robot with passive suction cups and self-locking trapezoidal lead screw drive

Thein Than Tun et al.

AUTOMATION IN CONSTRUCTION (2018)

Article Engineering, Electrical & Electronic

Design of sTetro: A Modular, Reconfigurable, and Autonomous Staircase Cleaning Robot

Muhammad Ilyas et al.

JOURNAL OF SENSORS (2018)

Article Computer Science, Artificial Intelligence

Planning Stable and Efficient Paths for Reconfigurable Robots On Uneven Terrain

Mohammad Norouzi et al.

JOURNAL OF INTELLIGENT & ROBOTIC SYSTEMS (2017)

Article Engineering, Electrical & Electronic

Autonomous Network Repairing of a Home Security System Using Modular Self-Reconfigurable Robots

Guifang Qiao et al.

IEEE TRANSACTIONS ON CONSUMER ELECTRONICS (2013)

Article Computer Science, Interdisciplinary Applications

Transformation Design Theory: A Meta-Analogical Framework

Jason Weaver et al.

JOURNAL OF COMPUTING AND INFORMATION SCIENCE IN ENGINEERING (2010)

Review Engineering, Mechanical

Innovations in Design Through Transformation: A Fundamental Study of Transformation Principles

Vikramjit Singh et al.

JOURNAL OF MECHANICAL DESIGN (2009)

Article Computer Science, Artificial Intelligence

A fast and elitist multiobjective genetic algorithm: NSGA-II

K Deb et al.

IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION (2002)