Corpus Curation for Cognitive Robots via Kinematic Motion Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for creating corpora for cognitive robots are inadequate as they limit robot responses and require extensive re-mapping for changes in gestures and movements, making them unsuitable for robot-based interactions and inefficient.
Innovation Solution
A corpus curation method that maps kinematic motions of robots to granular features of items in the corpus, allowing for consistent behavior across applications without the need for re-mapping responses, and enables offline updates of tags for kinematic motion delivery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If gestures and movements are mapped to individual responses or words, then robot responses can be customized, but the number of responses is limited and extensive re-mapping is required for changes
Solution Approach 1:
The patent segments the corpus into fine-grained features (such as object properties, actions, attributes) rather than mapping to entire responses or words. This segmentation allows independent modification of feature mappings without affecting other parts of the system, thereby reducing re-mapping complexity while maintaining response customization capability
Solution Approach 2:
The patent creates a universal feature-based mapping layer that serves multiple functions: it enables response customization, facilitates easy updates through offline corpus modifications, and works across different robot applications. The feature mapping system acts as an intermediary that can be updated without reprogramming individual response mappings
2Stability of the object's composition
If each response is re-mapped based on a change for one type of movement, then gesture consistency is maintained, but hundreds of thousands of changes are required
Solution Approach 1:
The patent performs preliminary action by creating an offline corpus with pre-defined fine-grained feature mappings before deployment. When gesture changes are needed, only the corpus features need updating rather than individual response mappings. This preliminary structuring of the corpus allows rapid adaptation while maintaining gesture consistency across all responses that share the same features
Solution Approach 2:
The patent introduces fine-grained features as an intermediary layer between the corpus content and the robot's kinematic motions. This intermediary abstraction allows changes to be made at the feature level rather than at the response level, significantly reducing the number of modifications required while maintaining consistency through the intermediary feature mappings
3Ease of manufacture
If programmer's intuition is used to link moves to corpus sections, then implementation is simple, but gestures are not tied to corpus content
Solution Approach 1:
The patent changes the parameters of corpus representation from coarse-grained sections to fine-grained features with specific attributes. This parameter change enables systematic and reliable linking of gestures to corpus content based on feature characteristics rather than programmer intuition, while the structured feature framework maintains implementation simplicity through automated mapping processes
Data Source
AI summary
A corpus curation method, system, and non-transitory computer readable medium, include mapping a kinematic motion of a robot to a granular feature of an item in the corpus and answering a user question using the mapped kinematic motion embedded in an answer by the robot.