Collision-Free Kinematic Posture Prediction Using CNN Depth Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for simulating collision-free postures for complex kinematic systems like robots and humans are time-consuming and require significant manual interaction or computational effort, limiting the efficiency of human simulation technology.
Innovation Solution
A method using a machine learning-based collision-free detection function (CFD) trained with a convolutional neural network (CNN) to predict collision-free postures by transforming the kinematic problem into an image classification problem, allowing for rapid prediction of postures that avoid collisions in cluttered environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional collision detection methods (RRT, voxelization) are used to find collision-free postures, then collision-free postures can be identified, but the computational time and processing cost increase significantly
Solution Approach 1:
The system pre-calculates and stores collision information in a depth map before posture prediction is needed. This depth map pre-processing creates a ready-to-use collision database that eliminates the need for time-consuming real-time collision detection during posture generation, directly resolving the contradiction between reliable collision-free identification and computational time consumption
Solution Approach 2:
The patent replaces traditional mechanical collision detection algorithms (RRT, voxelization) with a machine learning-based prediction system. The CFD function uses a depth map and kinematic parameters to directly predict collision-free postures, substituting complex computational mechanics with an optimized neural network approach that maintains accuracy while dramatically reducing processing time
2Reliability
If manual adjustment of joints is used to achieve collision-free postures, then collision-free postures can be obtained, but significant manual interaction time is required
Solution Approach 1:
The system implements self-service by automatically generating collision-free postures through the CFD function without requiring manual joint adjustment. The machine learning model independently processes the depth map and kinematic parameters to produce accurate collision-free postures, eliminating the time-consuming manual interaction while maintaining high reliability
Solution Approach 2:
Manual joint adjustment operations are replaced with an automated machine learning-based prediction system. The CFD function automatically computes optimal postures by processing depth map data and kinematic constraints, substituting labor-intensive manual operations with an efficient computational system that delivers reliable results without manual intervention
3Measurement precision
If comprehensive collision detection is performed on all postures, then accurate collision-free identification is achieved, but computational complexity and processing cost increase
Solution Approach 1:
The system extracts only the essential collision information from the environment and represents it compactly in a depth map. This extracted depth map contains precisely the collision data needed for prediction without including unnecessary environmental details, maintaining high identification accuracy while significantly reducing computational complexity compared to comprehensive collision detection
Solution Approach 2:
The patent changes the parameter representation from comprehensive 3D environment models to simplified depth map parameters. This parameter transformation reduces the dimensional complexity of the problem while preserving the critical collision information needed for accurate posture prediction, resolving the contradiction between measurement precision and device complexity
Data Source
AI summary
A system and a method predict a collision free posture of a kinematic system. The method includes: receiving a 3D virtual environment, receiving a 3D representation of the kinematic system and a set of 3D postures defined for the 3D virtual kinematic system, receiving a target task to be performed by the kinematic system with respect to the surrounding environment, and receiving a prescribed location within the 3D virtual environment. The prescribed location defines a position at which the 3D virtual kinematic system has to be placed within the 3D virtual environment. A collision free detection function (CFD) is applied to a set of input data containing the 3D virtual environment, the target task, the prescribed location and the set of postures. The CFD function outputs a set of collision free postures enabling the kinematic system to perform the target task when located at the prescribed location.