3D Object Convex Decomposition for Collision Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for simplifying 3D object representations for collision detection often result in sub-optimal performance due to false collision detection and high computational complexity.
Innovation Solution
A method of decomposing a three-dimensional representation of an object into a plurality of convex hulls by using a cluster priority queue, computing concavity measures, and applying cut planes to divide clusters into new convex hulls, while performing hierarchical searches and accepting user input to optimize the process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If simplified representation of 3D objects is used for collision detection, then computational complexity is reduced, but collision detection accuracy deteriorates
Solution Approach 1:
The patent divides a 3D object into multiple convex clusters by applying cut planes, creating a hierarchical decomposition. This segmentation allows the object to be represented by a set of convex hulls that can be processed more efficiently while maintaining collision detection accuracy through the priority queue ordering mechanism.
Solution Approach 2:
The patent changes the representation parameters from a single simplified shape to multiple convex clusters with varying levels of detail. By using a priority queue to order clusters by concavity measure, the system dynamically adjusts the level of detail used in collision detection, improving both accuracy and computational efficiency.
2Device complexity
If existing simplification techniques are applied, then computational complexity is reduced, but collision detection performance deteriorates due to false detection
Solution Approach 1:
The patent performs preliminary computation of concavity measures and hierarchical ordering of clusters before collision detection. By pre-computing the priority queue ordering based on concavity measures, the system prepares the data structure in advance, allowing for more accurate and efficient collision detection without increasing runtime computational complexity.
Solution Approach 2:
The patent incorporates feedback mechanisms through user input acceptance and iterative refinement. The system allows users to review and correct the decomposition, providing feedback that improves the quality of the convex cluster representation and reduces false collision detections while maintaining computational efficiency.
3Speed
If automated decomposition is used, then processing speed is improved, but control flexibility is reduced
Solution Approach 1:
The patent creates a dynamic system where automated decomposition can be adjusted based on user input. The priority queue and cluster ordering can be modified interactively, allowing the system to adapt between fully automated processing and user-guided refinement, providing both speed and flexibility.
Solution Approach 2:
The system performs self-service through automated concavity measurement and cluster ordering, but allows users to intervene when needed. The automated decomposition handles routine processing efficiently, while user input can be incorporated to refine specific aspects of the decomposition, combining the speed of automation with the flexibility of human control.
Data Source
AI summary
A method of decomposing a three-dimensional representation of an object into a plurality of convex hulls can include instantiating a cluster priority queue in a computing system memory that initially contains a cluster corresponding to the three-dimensional representation of the object, computing with a processor of the computing system a concavity measure for each cluster in the cluster priority queue, and, for the cluster with the highest concavity measure: (1) computing with the processor a cut plane that divides the cluster corresponding to the three-dimensional representation of the object into two new clusters, each of the two new clusters having a corresponding convex hull, wherein computing a cut plane includes performing a hierarchical search of potential cut planes, (2) removing the cluster corresponding to the three-dimensional representation of the object from the cluster priority queue, and (3) adding the two new clusters to the cluster priority queue.