Robotic Corner Detection for Grasping Unknown-Size Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Robotic systems face challenges in accurately identifying and interacting with objects when their dimensions and positions are unknown or known with incomplete accuracy, particularly when objects are positioned flush with each other, leading to difficulties in grasping and transporting objects correctly.
Innovation Solution
A computing system that uses image information from a camera to detect corners and define a minimum viable region for a target open corner, generating candidate edge segments, determining intersection points, and selecting a minimum viable region candidate to assist robotic systems in grasping objects without needing exact edge identification, allowing for adjustment based on updated dimensions post-grasping.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the robotic system attempts to identify exact edges and dimensions of objects before grasping, then the precision of object manipulation improves, but the time required for detection and the complexity of the system increase
Solution Approach 1:
The system performs preliminary corner detection and defines a minimum viable region before the grasping action. By detecting corners and generating candidate edge segments in advance, the system establishes a preliminary understanding of object boundaries without requiring complete edge identification, thus reducing detection time while maintaining sufficient precision for accurate grasping.
Solution Approach 2:
Instead of detecting all edges and dimensions of an object, the system performs partial detection by identifying only the minimum viable region that contains the object. This partial action approach detects corners and generates candidate regions without completing full edge detection, significantly reducing computational time and system complexity while providing enough information for successful grasping.
2Productivity
If the robotic system uses a simple detection method to quickly identify objects, then the processing speed improves, but the reliability of object identification deteriorates
Solution Approach 1:
The detection process is segmented into distinct stages: corner detection, candidate edge segment generation, intersection point calculation, and minimum viable region selection. Each segment performs a specific function with manageable complexity, allowing fast processing at each stage while ensuring reliable object identification through the cumulative effect of all segments working together.
Solution Approach 2:
The system incorporates validation feedback by checking whether the selected minimum viable region candidate actually contains an object corner. This feedback mechanism ensures that the fast detection process maintains reliability by verifying results at critical decision points, allowing the system to quickly identify objects while filtering out false detections.
3Reliability
If the robotic system defines a large detection region to ensure object coverage, then the reliability of object containment improves, but the precision of locating the actual object boundaries deteriorates
Solution Approach 1:
The system defines a minimum viable region that is larger than the actual object boundaries to ensure complete containment, but only detects corners and generates candidate segments for this partial region. This excessive action in region definition guarantees object containment while the partial detection of only corners and candidate edges maintains precision by avoiding unnecessary detection of the entire large region's boundaries.
Solution Approach 2:
The detection region is segmented into multiple candidate edge segments that are evaluated individually. By dividing the large detection region into smaller candidate segments and evaluating their intersections, the system maintains precision in boundary localization while ensuring the overall minimum viable region reliably contains the object through the collective coverage of all segments.
Data Source
AI summary
A computing system including a processing circuit in communication with a camera having a field of view. The processing circuit obtains image information based on the objects in the field of view and defines a minimum viable region for a target open corner. Potential minimum viable regions are defined by identifying candidate edges of an object and determining potential intersection points based on the candidate edges. The minimum viable region may then be identified and validated from the potential minimum viable regions.