Development of Technology for Robot Arms to Grasp Transparent and Glossy Objects Quickly and Accurately - Advancing Production Site Automation, Reducing Work Time, and Improving Productivity
Tokyo University of Science researchers developed an algorithm enabling robot arms to accurately grasp transparent/glossy objects with a 96% success rate using a single camera, significantly reducing operation time.
📋 Article Processing Timeline
- 📰 Published: March 30, 2026 at 19:00
- 🔍 Collected: March 30, 2026 at 22:56 (3h 56m after Published)
- 🤖 AI Analyzed: April 24, 2026 at 05:49 (582h 53m after Collected)
**[Summary and Points of the Research]**
We have developed a technology that estimates the shape of target objects that conventional 3D measurement struggles with, such as transparent containers and glossy packaging materials, from a single camera image, enabling a robot arm to grasp them.
Even when observation from multiple viewpoints is necessary, we developed a method to automatically determine the shooting position and movement path by balancing shape estimation accuracy with movement distance.
Verification with an actual robot achieved a grasping success rate of 96.0%, reducing the camera's moving distance by 52% and the overall handling execution time by 19% compared to conventional methods.
This research result is expected to promote the automation of processes that previously relied on human hands, contributing to improved productivity by balancing high-precision grasping with efficient operation.
**[Overview of the Research]**
A research group led by Associate Professor Shogo Arai of the Department of Mechanical and Aerospace Engineering, Faculty of Science and Technology, Tokyo University of Science (TUS), and Kenneth Ginga (2nd-year Master's student in 2025) at the Graduate School of Science and Technology, TUS, has developed '3D measurement (*1)' and 'grasp planning (*2)' methods for target objects that are difficult for robot arms to grasp, such as transparent or glossy objects. Furthermore, in an eye-in-hand configuration (*3), they succeeded in reducing the movement and processing time for taking images.
Conventionally, transparent containers and glossy packaging materials cause standard depth sensors and typical 3D measurements to become unstable due to light reflection and transmission on their surfaces, making automatic grasping by robot arms difficult. Therefore, the research group focused on a method that combines semantic segmentation (*4) of RGB images, which is less affected by optical properties, with 'Shape from Silhouette' (*5), which reconstructs shapes from multi-viewpoint contour information. However, while multi-viewpoint imaging improves accuracy, it takes time to move the camera, posing a challenge in balancing it with the tact time required on manufacturing floors. To solve this problem, they introduced a cost function that balances improving 3D measurement accuracy with shortening the camera movement distance, thereby optimizing the shooting positions and movement paths.
In validation using an actual robot, they achieved a grasp success rate of 96.0% for transparent, glossy, and opaque objects, successfully reducing the camera movement distance by 52% and the overall handling execution time by 19% compared to baseline methods.
This research achieved both robustness in grasping objects with difficult optical properties and a reduction in the time costs associated with multi-viewpoint observation. This broadens the scope of robot applications and is expected to contribute to promoting automation and improving productivity at manufacturing sites.
This research outcome was published online in the international academic journal 'IEEE ROBOTICS AND AUTOMATION LETTERS' on January 12, 2026. Moreover, this achievement is scheduled to be presented at the 2026 IEEE International Conference on Robotics & Automation (ICRA 2026), the top conference in the robotics field, garnering global attention.
Figure 1 Prio...
We have developed a technology that estimates the shape of target objects that conventional 3D measurement struggles with, such as transparent containers and glossy packaging materials, from a single camera image, enabling a robot arm to grasp them.
Even when observation from multiple viewpoints is necessary, we developed a method to automatically determine the shooting position and movement path by balancing shape estimation accuracy with movement distance.
Verification with an actual robot achieved a grasping success rate of 96.0%, reducing the camera's moving distance by 52% and the overall handling execution time by 19% compared to conventional methods.
This research result is expected to promote the automation of processes that previously relied on human hands, contributing to improved productivity by balancing high-precision grasping with efficient operation.
**[Overview of the Research]**
A research group led by Associate Professor Shogo Arai of the Department of Mechanical and Aerospace Engineering, Faculty of Science and Technology, Tokyo University of Science (TUS), and Kenneth Ginga (2nd-year Master's student in 2025) at the Graduate School of Science and Technology, TUS, has developed '3D measurement (*1)' and 'grasp planning (*2)' methods for target objects that are difficult for robot arms to grasp, such as transparent or glossy objects. Furthermore, in an eye-in-hand configuration (*3), they succeeded in reducing the movement and processing time for taking images.
Conventionally, transparent containers and glossy packaging materials cause standard depth sensors and typical 3D measurements to become unstable due to light reflection and transmission on their surfaces, making automatic grasping by robot arms difficult. Therefore, the research group focused on a method that combines semantic segmentation (*4) of RGB images, which is less affected by optical properties, with 'Shape from Silhouette' (*5), which reconstructs shapes from multi-viewpoint contour information. However, while multi-viewpoint imaging improves accuracy, it takes time to move the camera, posing a challenge in balancing it with the tact time required on manufacturing floors. To solve this problem, they introduced a cost function that balances improving 3D measurement accuracy with shortening the camera movement distance, thereby optimizing the shooting positions and movement paths.
In validation using an actual robot, they achieved a grasp success rate of 96.0% for transparent, glossy, and opaque objects, successfully reducing the camera movement distance by 52% and the overall handling execution time by 19% compared to baseline methods.
This research achieved both robustness in grasping objects with difficult optical properties and a reduction in the time costs associated with multi-viewpoint observation. This broadens the scope of robot applications and is expected to contribute to promoting automation and improving productivity at manufacturing sites.
This research outcome was published online in the international academic journal 'IEEE ROBOTICS AND AUTOMATION LETTERS' on January 12, 2026. Moreover, this achievement is scheduled to be presented at the 2026 IEEE International Conference on Robotics & Automation (ICRA 2026), the top conference in the robotics field, garnering global attention.
Figure 1 Prio...