On the design of deep learning-based control algorithms for visually guided UAVs engaged in power tower inspection tasks
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.Training Convolution Neural Networks with a hybrid dataset of synthetic and real images improves Unmanned Aerial Vehicle (UAV) inspection. This approach enhances image segmentation for autonomous power tower inspection, outperforming models trained on single data types.
Area Of Science
- Computer Vision
- Robotics
- Artificial Intelligence
Background
- Autonomous Unmanned Aerial Vehicles (UAVs) require robust visual guidance for tasks like power tower inspection.
- Image segmentation for UAVs is challenging due to varying tower structures and complex backgrounds.
- Manual data annotation for training is labor-intensive and costly.
Purpose Of The Study
- To design and evaluate Convolution Neural Networks (CNNs) for visually guiding UAVs in power tower inspection.
- To compare the performance of CNNs trained on synthetic, physical-world, and hybrid image datasets.
- To assess the effectiveness of photogrammetry in generating synthetic training data.
Main Methods
- Utilized an attention-based U-NET architecture for image segmentation.
- Generated synthetic images using photogrammetry and simulated UAV environments.
- Conducted a comparative study using synthetic, physical-world, and hybrid datasets for network training.
- Evaluated network performance using multiple image segmentation metrics.
Main Results
- The CNN trained on the hybrid dataset significantly outperformed those trained solely on synthetic or physical-world data.
- Hybrid dataset training demonstrated superior performance across various image segmentation evaluation metrics.
- Photogrammetry proved effective in creating valuable synthetic datasets for UAV inspection tasks.
Conclusions
- A hybrid approach combining synthetic and physical-world images offers the optimal balance between cost-efficiency and performance for training UAV inspection networks.
- The study validates the potential of using photogrammetry-generated data to automate precise UAV movements for infrastructure inspection.
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
Related Concept Videos
Proportional Integral (PI) controllers are a fundamental component in modern control systems, widely used to enhance performance and mitigate steady-state errors. They are particularly effective in applications such as automatic brightness adjustment on smartphones, where they excel at mitigating steady-state errors for step-function inputs. Unlike PD controllers, which require time-varying errors to function optimally, PI controllers leverage their integral component to address residual...
In automotive engineering, car suspension systems often employ Proportional Derivative (PD) controllers to enhance performance. PD controllers are utilized to adjust the damping force in response to road conditions. A controller, acting as an amplifier with a constant gain, demonstrates proportional control, with output directly mirroring input.
Designing a continuous-data controller requires selecting and linking components like adders and integrators, which are fundamental in Proportional,...