20230114734. METHOD AND APPARATUS WITH GLOBAL LOCALIZATION simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)
Contents
METHOD AND APPARATUS WITH GLOBAL LOCALIZATION
Organization Name
Inventor(s)
Hyewon Moon of Seongnam-si (KR)
Jiyeon Kim of Hwaseong-si (KR)
METHOD AND APPARATUS WITH GLOBAL LOCALIZATION - A simplified explanation of the abstract
This abstract first appeared for US patent application 20230114734 titled 'METHOD AND APPARATUS WITH GLOBAL LOCALIZATION
Simplified Explanation
The abstract describes a method for global localization using two neural networks. Here is a simplified explanation of the abstract:
- The method involves using two neural networks to estimate the location and pose of an input image.
- The first network extracts features from the input image.
- The second network estimates a coordinate map corresponding to the input image using the extracted features.
- The pose of the input image is then estimated based on the estimated coordinate map.
- The first and/or second network can be trained using generative adversarial network (GAN) loss.
- The GAN loss is determined based on features extracted from synthetic and real images, as well as coordinate maps estimated by the second network.
Potential applications of this technology:
- Autonomous vehicles: This method can be used for accurate localization of vehicles, enabling them to navigate and make decisions based on their precise location.
- Robotics: Robots can use this method to determine their position in an environment, allowing them to perform tasks more effectively.
- Augmented reality: This technology can be used to accurately overlay virtual objects onto the real world, enhancing the user's experience.
Problems solved by this technology:
- Accurate global localization: The method provides a reliable way to estimate the location and pose of an input image, even in complex and changing environments.
- Robustness to different image types: By training the networks using both synthetic and real images, the method can handle variations in image quality and appearance.
Benefits of this technology:
- Improved navigation and decision-making: Accurate global localization allows vehicles and robots to navigate and make decisions based on their precise location, leading to safer and more efficient operations.
- Enhanced user experience: Augmented reality applications can benefit from accurate localization, providing users with more realistic and immersive experiences.
- Versatility: The method can handle different types of images, making it applicable to various domains and scenarios.
Original Abstract Submitted
a method with global localization includes: extracting a feature by applying an input image to a first network; estimating a coordinate map corresponding to the input image by applying the extracted feature to a second network; and estimating a pose corresponding to the input image based on the estimated coordinate map, wherein either one or both of the first network and the second network is trained based on either one or both of: a first generative adversarial network (gan) loss determined based on a first feature extracted by the first network based on a synthetic image determined by three-dimensional (3d) map data and a second feature extracted by the first network based on a real image; and a second gan loss determined based on a first coordinate map estimated by the second network based on the first feature and a second coordinate map estimated by the second network based on the second feature.
