18463559. COMPUTER-IMPLEMENTED METHOD FOR GENERATING (TRAINING) IMAGES simplified abstract (Robert Bosch GmbH)
Contents
COMPUTER-IMPLEMENTED METHOD FOR GENERATING (TRAINING) IMAGES
Organization Name
Inventor(s)
Yoel Shapiro of Kiryat Bialik (IL)
COMPUTER-IMPLEMENTED METHOD FOR GENERATING (TRAINING) IMAGES - A simplified explanation of the abstract
This abstract first appeared for US patent application 18463559 titled 'COMPUTER-IMPLEMENTED METHOD FOR GENERATING (TRAINING) IMAGES
Simplified Explanation
The computer-implemented method described in the abstract involves generating labelled training images that characterize the manipulation of stackable objects in a workspace. Here are the key points of the innovation:
- Obtaining a first training image subset with a depth map and visual image of stackable objects in a stacking region at a first time index.
- Obtaining a second training image subset with a depth map and visual image of the stacking region at a second time index, showing a changed spatial state.
- Computing depth and visual difference masks based on the images to identify changes.
- Generating an annotated segmentation mask using the difference masks to highlight the changes in the stacking region.
- Potential Applications
This technology can be applied in robotics for object manipulation tasks, automated quality control in manufacturing, and augmented reality applications for interactive experiences.
- Problems Solved
This technology solves the problem of accurately detecting and characterizing changes in the spatial state of stackable objects in a workspace, which is crucial for tasks like object manipulation and quality control.
- Benefits
The benefits of this technology include improved efficiency in object manipulation tasks, enhanced quality control processes, and the ability to create interactive and engaging augmented reality experiences.
- Potential Commercial Applications
Potential commercial applications of this technology include robotics companies for automation solutions, manufacturing industries for quality control systems, and entertainment companies for creating interactive AR experiences.
- Possible Prior Art
One possible prior art in this field could be systems that use computer vision and depth sensing technologies for object detection and tracking in various applications. However, the specific method described in this patent application appears to be novel in its approach to characterizing manipulation of stackable objects in a workspace.
- Unanswered Questions
- How does this technology handle occlusions or partial visibility of stackable objects in the workspace?
The abstract does not provide details on how the method deals with occlusions or partial visibility of stackable objects, which could be crucial in real-world scenarios where objects may be partially hidden from view.
- What is the computational complexity of the method for generating annotated segmentation masks, and how does it scale with the number of stackable objects in the workspace?
The abstract does not mention the computational complexity of the method or how it scales with the number of stackable objects, which could be important for assessing the feasibility of deploying this technology in real-time applications with multiple objects.
Original Abstract Submitted
A computer-implemented method for generating labelled training images characterizing manipulation of a plurality of stackable objects in a workspace. The method includes: obtaining a first training image subset obtained at a first time index comprising a depth map and a visual image of a plurality of stackable objects in a stacking region of a workspace; obtaining a second training image subset obtained at a second time index comprising a depth map and a visual image of the stacking region in the workspace, wherein the second training image subset characterizes a changed spatial state of the stacking region; computing a depth difference mask based on the depth maps; computing a visual difference mask based on the visual images of the first and second training image subsets; generating an annotated segmentation mask using the depth difference mask and/or the visual difference mask.
