Kenji Shimada
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Kenji Shimada has a total of 106 co-authors. Here are the top co-authors by number of shared publications:
- In-Ho Song
Carnegie Mellon University
5 shared publications - Jonathan Cagan
Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, USA
5 shared publications - Soji Yamakawa
Carnegie Mellon University Department of Mechanical Engineering Pittsburgh, PA 15213
5 shared publications - Yoed Rabin
Biothermal Technology Laboratory, Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA, 15213, USA
5 shared publications Hong Zhang
Department of Mechanical Engineering, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, USA
4 shared publicationsSoji Yamakawa
Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA, U.S.A.
4 shared publications
Latest Publications
- 26 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (11) , 11586-11593
Concrete Spray 3D Printing Simulator for Nozzle Trajectory Planning
Existing 3D printing simulators target extrusion-based methods, which cannot be directly applied to spray-based methods. This paper presents a new computational method for simulating concrete spray 3D printing, where nozzle position, orientation, and speed are robotically controlled. One of the main benefits of concrete spray 3D printing is the ability to create an arbitrary shape that can support heavy loading and spray over reinforcing bars, or rebar, to create structurally sound architecture. The proposed simulator applies a voxel-based representation of substrate and sprayed concrete, with a novel application of mathematical modeling for concrete dripping, concrete particle rebounding and spread due to airflow, and a semi-transparency parameter to accurately emulate the spray's flow around rebars. Various physical parameters are also considered in the proposed method including nozzle design, material properties of the concrete mixture, spray apex angle, material rebound, etc. The simulator outputs the mass distribution of the concrete in voxels. Two test cases were performed with a real concrete spray 3D printing robot and replicated in simulation to compare. One test case focuses on the simulator's capabilities to predict shape based on nozzle motion speeds, producing an average accuracy of $90.75\%$ in height of sprayed concrete. The second test case showed the simulator's capabilities to predict printing over a rebar network, resulting in an average accuracy of $92.3\%$ and $97.9\%$ for width and thickness, respectively.
- 26 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (11) , 11586-11593
Concrete Spray 3D Printing Simulator for Nozzle Trajectory Planning
Existing 3D printing simulators target extrusion-based methods, which cannot be directly applied to spray-based methods. This paper presents a new computational method for simulating concrete spray 3D printing, where nozzle position, orientation, and speed are robotically controlled. One of the main benefits of concrete spray 3D printing is the ability to create an arbitrary shape that can support heavy loading and spray over reinforcing bars, or rebar, to create structurally sound architecture. The proposed simulator applies a voxel-based representation of substrate and sprayed concrete, with a novel application of mathematical modeling for concrete dripping, concrete particle rebounding and spread due to airflow, and a semi-transparency parameter to accurately emulate the spray's flow around rebars. Various physical parameters are also considered in the proposed method including nozzle design, material properties of the concrete mixture, spray apex angle, material rebound, etc. The simulator outputs the mass distribution of the concrete in voxels. Two test cases were performed with a real concrete spray 3D printing robot and replicated in simulation to compare. One test case focuses on the simulator's capabilities to predict shape based on nozzle motion speeds, producing an average accuracy of $90.75\%$ in height of sprayed concrete. The second test case showed the simulator's capabilities to predict printing over a rebar network, resulting in an average accuracy of $92.3\%$ and $97.9\%$ for width and thickness, respectively.
- 11 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (11) , 11196-11203
Semantic Exploration and Dense Mapping of Complex Environments Using Ground Robot With Panoramic LiDAR-Camera Fusion
This paper presents a system for autonomous semantic exploration and dense semantic target mapping of a complex unknown environment using a ground robot equipped with a panoramic LiDAR-camera system. Existing approaches often struggle to strike a balance between collecting enough high-quality observations from multiple view angles and avoiding unnecessary repetitive traversal. To fill this gap, we propose a complete system that combines mapping and planning. We first redefine the task as completing both geometric coverage and semantic viewpoint observation. Subsequently, we manage semantic and geometric viewpoints separately and propose a novel Priority-driven Decoupled Local Sampler to generate local viewpoint sets. This allows for explicit multi-view semantic inspection and voxel coverage without unnecessary repetition. Building on this, we develop a hierarchical planner that ensures efficient global coverage. In addition, we propose a Safe Aggressive Exploration State Machine, which allows the robot to extend its exploration path planning into unknown areas while ensuring the robot's safety with recovery behavior and adaptive sampling. Our system includes a modular semantic target mapping component designed to utilize odometry and point clouds from existing SLAM algorithms, enabling point-cloud-level dense semantic target mapping. We validate our approach through extensive experiments in both realistic simulations and complex real-world environments. Simulation results demonstrate that our planner achieves faster exploration and shorter travel distances while guaranteeing a specified number of multi-view inspections. Real-world experiments further confirm the effectiveness of the system in achieving accurate dense semantic object mapping of unstructured environments.
- 11 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (11) , 11196-11203
Semantic Exploration and Dense Mapping of Complex Environments Using Ground Robot With Panoramic LiDAR-Camera Fusion
This paper presents a system for autonomous semantic exploration and dense semantic target mapping of a complex unknown environment using a ground robot equipped with a panoramic LiDAR-camera system. Existing approaches often struggle to strike a balance between collecting enough high-quality observations from multiple view angles and avoiding unnecessary repetitive traversal. To fill this gap, we propose a complete system that combines mapping and planning. We first redefine the task as completing both geometric coverage and semantic viewpoint observation. Subsequently, we manage semantic and geometric viewpoints separately and propose a novel Priority-driven Decoupled Local Sampler to generate local viewpoint sets. This allows for explicit multi-view semantic inspection and voxel coverage without unnecessary repetition. Building on this, we develop a hierarchical planner that ensures efficient global coverage. In addition, we propose a Safe Aggressive Exploration State Machine, which allows the robot to extend its exploration path planning into unknown areas while ensuring the robot's safety with recovery behavior and adaptive sampling. Our system includes a modular semantic target mapping component designed to utilize odometry and point clouds from existing SLAM algorithms, enabling point-cloud-level dense semantic target mapping. We validate our approach through extensive experiments in both realistic simulations and complex real-world environments. Simulation results demonstrate that our planner achieves faster exploration and shorter travel distances while guaranteeing a specified number of multi-view inspections. Real-world experiments further confirm the effectiveness of the system in achieving accurate dense semantic object mapping of unstructured environments.
- 28 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (5) , 4946-4953
Intent Prediction-Driven Model Predictive Control for UAV Planning and Navigation in Dynamic Environments
Aerial robots can enhance construction site productivity by autonomously handling inspection and mapping tasks. However, ensuring safe navigation near human workers remains challenging. While navigation in static environments has been well studied, navigating dynamic environments remains open due to challenges in perception and planning. Payload limitations restrict the robots to using cameras with limited fields of view, resulting in unreliable perception and tracking during collision avoidance. Moreover, the rapidly changing conditions of dynamic environments can quickly make the generated optimal trajectory outdated.To address these challenges, this paper presents a comprehensive navigation framework that integrates perception, intent prediction, and planning. Our perception module detects and tracks dynamic obstacles efficiently and handles tracking loss and occlusion during collision avoidance. The proposed intent prediction module employs a Markov Decision Process (MDP) to forecast potential actions of dynamic obstacles with the possible future trajectories. Finally, a novel intent-based planning algorithm, leveraging model predictive control (MPC), is applied to generate navigation trajectories. Simulation and physical experiments11Experiment video link: https://youtu.be/4xsEeMB9WPY demonstrate that our method improves the safety of navigation by achieving the fewest collisions compared to benchmarks.
- 28 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (5) , 4946-4953
Intent Prediction-Driven Model Predictive Control for UAV Planning and Navigation in Dynamic Environments
Aerial robots can enhance construction site productivity by autonomously handling inspection and mapping tasks. However, ensuring safe navigation near human workers remains challenging. While navigation in static environments has been well studied, navigating dynamic environments remains open due to challenges in perception and planning. Payload limitations restrict the robots to using cameras with limited fields of view, resulting in unreliable perception and tracking during collision avoidance. Moreover, the rapidly changing conditions of dynamic environments can quickly make the generated optimal trajectory outdated.To address these challenges, this paper presents a comprehensive navigation framework that integrates perception, intent prediction, and planning. Our perception module detects and tracks dynamic obstacles efficiently and handles tracking loss and occlusion during collision avoidance. The proposed intent prediction module employs a Markov Decision Process (MDP) to forecast potential actions of dynamic obstacles with the possible future trajectories. Finally, a novel intent-based planning algorithm, leveraging model predictive control (MPC), is applied to generate navigation trajectories. Simulation and physical experiments11Experiment video link: https://youtu.be/4xsEeMB9WPY demonstrate that our method improves the safety of navigation by achieving the fewest collisions compared to benchmarks.
- Top Cited Papers
- 26 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (4) , 3668-3675
NavRL: Learning Safe Flight in Dynamic Environments
Safe flight in dynamic environments requires unmanned aerial vehicles (UAVs) to make effective decisions when navigating cluttered spaces with moving obstacles. Traditional approaches often decompose decision-making into hierarchical modules for prediction and planning. Although these handcrafted systems can perform well in specific settings, they might fail if environmental conditions change and often require careful parameter tuning. Additionally, their solutions could be suboptimal due to the use of inaccurate mathematical model assumptions and simplifications aimed at achieving computational efficiency. To overcome these limitations, this letter introduces the NavRL framework, a deep reinforcement learning-based navigation method built on the Proximal Policy Optimization (PPO) algorithm. NavRL utilizes our carefully designed state and action representations, allowing the learned policy to make safe decisions in the presence of both static and dynamic obstacles, with zero-shot transfer from simulation to real-world flight. Furthermore, the proposed method adopts a simple but effective safety shield for the trained policy, inspired by the concept of velocity obstacles, to mitigate potential failures associated with the black-box nature of neural networks. To accelerate the convergence, we implement the training pipeline using NVIDIA Isaac Sim, enabling parallel training with thousands of quadcopters. Simulation and physical experiments show that our method ensures safe navigation in dynamic environments and results in the fewest collisions compared to benchmarks.
- 26 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (4) , 3668-3675
Top Cited PapersNavRL: Learning Safe Flight in Dynamic Environments
Safe flight in dynamic environments requires unmanned aerial vehicles (UAVs) to make effective decisions when navigating cluttered spaces with moving obstacles. Traditional approaches often decompose decision-making into hierarchical modules for prediction and planning. Although these handcrafted systems can perform well in specific settings, they might fail if environmental conditions change and often require careful parameter tuning. Additionally, their solutions could be suboptimal due to the use of inaccurate mathematical model assumptions and simplifications aimed at achieving computational efficiency. To overcome these limitations, this letter introduces the NavRL framework, a deep reinforcement learning-based navigation method built on the Proximal Policy Optimization (PPO) algorithm. NavRL utilizes our carefully designed state and action representations, allowing the learned policy to make safe decisions in the presence of both static and dynamic obstacles, with zero-shot transfer from simulation to real-world flight. Furthermore, the proposed method adopts a simple but effective safety shield for the trained policy, inspired by the concept of velocity obstacles, to mitigate potential failures associated with the black-box nature of neural networks. To accelerate the convergence, we implement the training pipeline using NVIDIA Isaac Sim, enabling parallel training with thousands of quadcopters. Simulation and physical experiments show that our method ensures safe navigation in dynamic environments and results in the fewest collisions compared to benchmarks.
- 1 April 2024
- Published by Taylor & Francis in Nuclear Technology
- Vol. 210 (11) , 2204-2214
Determination of Gain Scheduling Parameters for Loss of a Feedwater Pump Transient Mitigation Using Neural Networks
I. I.A. Control systems for nuclear power plant processes are not regularly updated after their initial commissioning, except during major projects for component replacement or plant uprating. In o...
- 1 April 2024
- Published by Taylor & Francis in Nuclear Technology
- Vol. 210 (11) , 2204-2214
Determination of Gain Scheduling Parameters for Loss of a Feedwater Pump Transient Mitigation Using Neural Networks
I. I.A. Control systems for nuclear power plant processes are not regularly updated after their initial commissioning, except during major projects for component replacement or plant uprating. In o...
- Open Access
A real-time dynamic obstacle tracking and mapping system for UAV navigation and collision avoidance with an RGB-D camera
The real-time dynamic environment perception has become vital for autonomous robots in crowded spaces. Although the popular voxel-based mapping methods can efficiently represent 3D obstacles with arbitrarily complex shapes, they can hardly distinguish between static and dynamic obstacles, leading to the limited performance of obstacle avoidance. While plenty of sophisticated learning-based dynamic obstacle detection algorithms exist in autonomous driving, the quadcopter's limited computation resources cannot achieve real-time performance using those approaches. To address these issues, we propose a real-time dynamic obstacle tracking and mapping system for quadcopter obstacle avoidance using an RGB-D camera. The proposed system first utilizes a depth image with an occupancy voxel map to generate potential dynamic obstacle regions as proposals. With the obstacle region proposals, the Kalman filter and our continuity filter are applied to track each dynamic obstacle. Finally, the environment-aware trajectory prediction method is proposed based on the Markov chain using the states of tracked dynamic obstacles. We implemented the proposed system with our custom quadcopter and navigation planner. The simulation and physical experiments show that our methods can successfully track and represent obstacles in dynamic environments in real-time and safely avoid obstacles. Our software is available on GitHub as an open-source ROS package.
Open AccessA real-time dynamic obstacle tracking and mapping system for UAV navigation and collision avoidance with an RGB-D camera
The real-time dynamic environment perception has become vital for autonomous robots in crowded spaces. Although the popular voxel-based mapping methods can efficiently represent 3D obstacles with arbitrarily complex shapes, they can hardly distinguish between static and dynamic obstacles, leading to the limited performance of obstacle avoidance. While plenty of sophisticated learning-based dynamic obstacle detection algorithms exist in autonomous driving, the quadcopter's limited computation resources cannot achieve real-time performance using those approaches. To address these issues, we propose a real-time dynamic obstacle tracking and mapping system for quadcopter obstacle avoidance using an RGB-D camera. The proposed system first utilizes a depth image with an occupancy voxel map to generate potential dynamic obstacle regions as proposals. With the obstacle region proposals, the Kalman filter and our continuity filter are applied to track each dynamic obstacle. Finally, the environment-aware trajectory prediction method is proposed based on the Markov chain using the states of tracked dynamic obstacles. We implemented the proposed system with our custom quadcopter and navigation planner. The simulation and physical experiments show that our methods can successfully track and represent obstacles in dynamic environments in real-time and safely avoid obstacles. Our software is available on GitHub as an open-source ROS package.
- Open Access
Image-Enhanced U-Net: Optimizing Defect Detection in Window Frames for Construction Quality Inspection
Ensuring the structural integrity of window frames and detecting subtle defects, such as dents and scratches, is crucial for maintaining product quality. Traditional machine vision systems face challenges in defect identification, especially with reflective materials and varied environments. Modern machine and deep learning (DL) systems hold promise for post-installation inspections but face limitations due to data scarcity and environmental variability. Our study introduces an innovative approach to enhance DL-based defect detection, even with limited data. We present a comprehensive window frame defect detection framework incorporating optimized image enhancement, data augmentation, and a core U-Net model. We constructed five datasets using cell phones and the Spot Robot for autonomous inspection, evaluating our approach across various scenarios and lighting conditions in real-world window frame inspections. Our results demonstrate significant performance improvements over the standard U-Net model, with a notable 7.43% increase in the F1 score and 15.1% in IoU. Our approach enhances defect detection capabilities, even in challenging real-world conditions. To enhance the generalizability of this study, it would be advantageous to apply its methodology across a broader range of diverse construction sites.
Open AccessImage-Enhanced U-Net: Optimizing Defect Detection in Window Frames for Construction Quality Inspection
Ensuring the structural integrity of window frames and detecting subtle defects, such as dents and scratches, is crucial for maintaining product quality. Traditional machine vision systems face challenges in defect identification, especially with reflective materials and varied environments. Modern machine and deep learning (DL) systems hold promise for post-installation inspections but face limitations due to data scarcity and environmental variability. Our study introduces an innovative approach to enhance DL-based defect detection, even with limited data. We present a comprehensive window frame defect detection framework incorporating optimized image enhancement, data augmentation, and a core U-Net model. We constructed five datasets using cell phones and the Spot Robot for autonomous inspection, evaluating our approach across various scenarios and lighting conditions in real-world window frame inspections. Our results demonstrate significant performance improvements over the standard U-Net model, with a notable 7.43% increase in the F1 score and 15.1% in IoU. Our approach enhances defect detection capabilities, even in challenging real-world conditions. To enhance the generalizability of this study, it would be advantageous to apply its methodology across a broader range of diverse construction sites.
- 20 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (1) , 651-658
Onboard Dynamic-Object Detection and Tracking for Autonomous Robot Navigation With RGB-D Camera
Deploying autonomous robots in crowded indoor environments usually requires them to have accurate dynamic obstacle perception. Although plenty of previous works in the autonomous driving field have investigated the 3D object detection problem, the usage of dense point clouds from a heavy Light Detection and Ranging (LiDAR) sensor and their high computation cost for learning-based data processing make those methods not applicable to small robots, such as vision-based UAVs with small onboard computers. To address this issue, we propose a lightweight 3D dynamic obstacle detection and tracking (DODT) method based on an RGB-D camera, which is designed for low-power robots with limited computing power. Our method adopts a novel ensemble detection strategy, combining multiple computationally efficient but low-accuracy detectors to achieve real-time high-accuracy obstacle detection. Besides, we introduce a new feature-based data association and tracking method to prevent mismatches utilizing point clouds' statistical features. In addition, our system includes an optional and auxiliary learning-based module to enhance the obstacle detection range and dynamic obstacle identification. The proposed method is implemented in a small quadcopter, and the results show that our method can achieve the lowest position error (0.11 m) and a comparable velocity error (0.23 m/s) across the benchmarking algorithms running on the robot's onboard computer. The flight experiments prove that the tracking results from the proposed method can make the robot efficiently alter its trajectory for navigating dynamic environments. Our software is available on GitHub 1 as an open-source ROS package.
- 20 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (1) , 651-658
Onboard Dynamic-Object Detection and Tracking for Autonomous Robot Navigation With RGB-D Camera
Deploying autonomous robots in crowded indoor environments usually requires them to have accurate dynamic obstacle perception. Although plenty of previous works in the autonomous driving field have investigated the 3D object detection problem, the usage of dense point clouds from a heavy Light Detection and Ranging (LiDAR) sensor and their high computation cost for learning-based data processing make those methods not applicable to small robots, such as vision-based UAVs with small onboard computers. To address this issue, we propose a lightweight 3D dynamic obstacle detection and tracking (DODT) method based on an RGB-D camera, which is designed for low-power robots with limited computing power. Our method adopts a novel ensemble detection strategy, combining multiple computationally efficient but low-accuracy detectors to achieve real-time high-accuracy obstacle detection. Besides, we introduce a new feature-based data association and tracking method to prevent mismatches utilizing point clouds' statistical features. In addition, our system includes an optional and auxiliary learning-based module to enhance the obstacle detection range and dynamic obstacle identification. The proposed method is implemented in a small quadcopter, and the results show that our method can achieve the lowest position error (0.11 m) and a comparable velocity error (0.23 m/s) across the benchmarking algorithms running on the robot's onboard computer. The flight experiments prove that the tracking results from the proposed method can make the robot efficiently alter its trajectory for navigating dynamic environments. Our software is available on GitHub 1 as an open-source ROS package.
- Open Access
Improving Deep Learning-based Defect Detection on Window Frames with Image Processing Strategies
Detecting subtle defects in window frames, including dents and scratches, is vital for upholding product integrity and sustaining a positive brand perception. Conventional machine vision systems often struggle to identify these defects in challenging environments like construction sites. In contrast, modern vision systems leveraging machine and deep learning (DL) are emerging as potent tools, particularly for cosmetic inspections. However, the promise of DL is yet to be fully realized. A few manufacturers have established a clear strategy for AI integration in quality inspection, hindered mainly by issues like scarce clean datasets and environmental changes that compromise model accuracy. Addressing these challenges, our study presents an innovative approach that amplifies defect detection in DL models, even with constrained data resources. The paper proposes a new defect detection pipeline called InspectNet (IPT-enhanced UNET) that includes the best combination of image enhancement and augmentation techniques for pre-processing the dataset and a Unet model tuned for window frame defect detection and segmentation. Experiments were carried out using a Spot Robot doing window frame inspections . 16 variations of the dataset were constructed using different image augmentation settings. Results of the experiments revealed that, on average, across all proposed evaluation measures, Unet outperformed all other algorithms when IPT-enhanced augmentations were applied. In particular, when using the best dataset, the average Intersection over Union (IoU) values achieved were IPT-enhanced Unet, reaching 0.91 of mIoU.
Open AccessImproving Deep Learning-based Defect Detection on Window Frames with Image Processing Strategies
Detecting subtle defects in window frames, including dents and scratches, is vital for upholding product integrity and sustaining a positive brand perception. Conventional machine vision systems often struggle to identify these defects in challenging environments like construction sites. In contrast, modern vision systems leveraging machine and deep learning (DL) are emerging as potent tools, particularly for cosmetic inspections. However, the promise of DL is yet to be fully realized. A few manufacturers have established a clear strategy for AI integration in quality inspection, hindered mainly by issues like scarce clean datasets and environmental changes that compromise model accuracy. Addressing these challenges, our study presents an innovative approach that amplifies defect detection in DL models, even with constrained data resources. The paper proposes a new defect detection pipeline called InspectNet (IPT-enhanced UNET) that includes the best combination of image enhancement and augmentation techniques for pre-processing the dataset and a Unet model tuned for window frame defect detection and segmentation. Experiments were carried out using a Spot Robot doing window frame inspections . 16 variations of the dataset were constructed using different image augmentation settings. Results of the experiments revealed that, on average, across all proposed evaluation measures, Unet outperformed all other algorithms when IPT-enhanced augmentations were applied. In particular, when using the best dataset, the average Intersection over Union (IoU) values achieved were IPT-enhanced Unet, reaching 0.91 of mIoU.
- 13 June 2022
- Published by Wiley in Journal of Mechanical Design
- Vol. 144 (10) , 1-15
Three-Dimensional Ship Hull Encoding and Optimization via Deep Neural Networks
Design and optimization of hull shapes for optimal hydrodynamic performance have been a major challenge for naval architectures. Deep learning bears the promise of comprehensive geometric representation and new design synthesis. In this work, we develop a deep neural network (DNN)-based approach to encode the hull designs to condensed representations, synthesize novel designs, and optimize the synthetic design based on the hydrodynamic performance. A variational autoencoder (VAE) with the hydro-predictor is developed to learn the representation through reconstructing the Laplacian parameterized hulls and encode the geometry-drag function simulated through computational fluid dynamics (CFD). Two data augmentation techniques, Perlin noise mapping and free-form deformation (FFD), are implemented to create the training set from a parent hull. The trained VAE is leveraged to efficiently optimize from massive synthetic hull vessels toward the optimal predicted drag performance. The selected geometries are further investigated and virtually screened under CFD simulations. Experiments show that our convolutional neural network (CNN) model accurately reconstructs the input vessels and predicts the corresponding drag coefficients. The proposed framework is demonstrated to synthesize realistic hull designs and optimize toward new hull designs with the drag coefficient decreased by 35% comparing to the parent design.
- 13 June 2022
- Published by Wiley in Journal of Mechanical Design
- Vol. 144 (10) , 1-15
Three-Dimensional Ship Hull Encoding and Optimization via Deep Neural Networks
Design and optimization of hull shapes for optimal hydrodynamic performance have been a major challenge for naval architectures. Deep learning bears the promise of comprehensive geometric representation and new design synthesis. In this work, we develop a deep neural network (DNN)-based approach to encode the hull designs to condensed representations, synthesize novel designs, and optimize the synthetic design based on the hydrodynamic performance. A variational autoencoder (VAE) with the hydro-predictor is developed to learn the representation through reconstructing the Laplacian parameterized hulls and encode the geometry-drag function simulated through computational fluid dynamics (CFD). Two data augmentation techniques, Perlin noise mapping and free-form deformation (FFD), are implemented to create the training set from a parent hull. The trained VAE is leveraged to efficiently optimize from massive synthetic hull vessels toward the optimal predicted drag performance. The selected geometries are further investigated and virtually screened under CFD simulations. Experiments show that our convolutional neural network (CNN) model accurately reconstructs the input vessels and predicts the corresponding drag coefficients. The proposed framework is demonstrated to synthesize realistic hull designs and optimize toward new hull designs with the drag coefficient decreased by 35% comparing to the parent design.