Object Estimation Apparatus Using Restricted Range Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for extracting piled objects face inefficiencies in estimating the position and orientation of workpieces, particularly when the piled state changes, as they require extensive processing time and struggle to maintain estimation accuracy.
Innovation Solution
An information processing apparatus and method that estimates the state of objects using image data, selects objects for extraction based on initial state information, and generates restriction range information to optimize subsequent estimation processing, thereby reducing processing time and maintaining accuracy even when the piled state changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full estimation processing is performed on all objects to ensure accuracy, then estimation accuracy is maintained, but processing time increases
Solution Approach 1:
The patent segments the estimation processing into two distinct stages: initial estimation processing for all objects to establish baseline state information, and subsequent restricted estimation processing that applies pre-determined ranges to limit the search space. This segmentation allows the system to maintain accuracy while reducing processing time in repeated operations.
Solution Approach 2:
The patent performs preliminary estimation processing on all objects before extraction to obtain initial state information, including position, orientation, and size. This preliminary action establishes reference data that is used to create restricted ranges for subsequent estimation, eliminating the need to perform full estimation processing on every object and thereby reducing processing time while maintaining accuracy.
2Productivity
If restriction ranges are applied to reduce processing time, then processing speed increases, but estimation accuracy may deteriorate when piled state changes
Solution Approach 1:
The patent implements a feedback mechanism where the actual position and orientation of extracted objects are compared with the estimated values. This feedback information is used to update and refine the restricted ranges for subsequent estimation processing, ensuring that even when piled state changes occur, the estimation remains accurate by adapting the ranges based on observed deviations.
Solution Approach 2:
The patent makes the restriction ranges dynamic rather than static. The ranges are initially determined based on pre-extraction state information but are subsequently updated and adjusted based on actual extraction results and changes in the piled state. This dynamic adjustment allows the system to maintain accuracy while benefiting from the speed advantages of restricted processing.
3Productivity
If association between extracted object and next object is established to reuse estimation data, then processing efficiency improves, but reliability decreases when relationship is tenuous
Solution Approach 1:
The patent changes the parameter representation from object identity associations to spatial and geometric parameter associations. Instead of relying on relationships between specific objects, the system uses state information (position, orientation, size) and pre-determined ranges that are applicable to any object in the pile. This parameter-based approach maintains reliability even when object relationships are tenuous, while still enabling efficient reuse of estimation data through the restricted ranges.
Data Source
AI summary
An information processing apparatus includes an estimation unit configured to estimate a state of a plurality of objects based on image data obtained by capturing the plurality of object with an imaging unit, a selection unit configured to select an object to be extracted from the plurality of objects based on first state information that indicates the state of the plurality of objects estimated by the estimation unit, and a generation unit configured to, based on second state information that indicates a state of an object other than the selected object among the plurality of objects, generate restriction range information that indicates a range for restricting estimation processing on a state of the other object, wherein the estimation unit estimates the state of the other object based on the restriction range information.