Robot Path Planning Graphs With Partial-Pose Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-dimensional configuration spaces for robots with many degrees of freedom pose significant computational challenges in online path planning, leading to inefficient graph generation and low-quality motion paths.
Innovation Solution
A method for generating graphs for online path planning that involves unconstrained sampling of partial robot poses, followed by constrained determination of remaining configuration parameters using a distance function related to reference robot poses, reducing dimensionality and ensuring high-quality motion paths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If sampling-based graph generation is used for high-DoF robots, then path planning can be performed, but the computational burden becomes excessive due to the enormous search space in high-dimensional configuration space
Solution Approach 1:
The patent segments the configuration space sampling process into two distinct phases: (1) unconstrained sampling of reduced-dimensional partial poses to generate candidate nodes, and (2) constrained refinement of remaining configuration parameters using distance functions. This segmentation allows the system to explore the configuration space efficiently without being overwhelmed by the full dimensionality, thereby reducing computational time while maintaining path planning capability
Solution Approach 2:
The patent applies dimensionality reduction by sampling only a subset of configuration parameters (partial poses) rather than the full configuration space. By working in a reduced dimension and then refining the remaining parameters through constrained optimization using distance functions to reference poses, the system effectively transforms the high-dimensional sampling problem into a lower-dimensional problem, significantly reducing the computational burden
2Manufacturing precision
If the graph contains sufficiently large and distributed collection of nodes to provide smooth high-quality motion, then path quality improves, but the computational burden of graph generation increases significantly
Solution Approach 1:
The patent performs preliminary unconstrained sampling of reduced-dimensional partial poses to generate a set of candidate graph nodes before refining their complete configuration parameters. This preliminary action creates a distributed set of candidate positions that ensure good spatial coverage and motion quality, while the subsequent constrained refinement step completes the configuration parameters efficiently using distance functions to reference poses, avoiding the need to generate and refine all possible configurations
Solution Approach 2:
The patent samples only a partial set of configuration parameters (partial poses) rather than the complete configuration space, generating sufficient candidate nodes for high-quality motion planning without exhaustively exploring all dimensions. This partial sampling approach, combined with constrained refinement of the remaining parameters, achieves the necessary node distribution for smooth motion while keeping graph generation computationally feasible
Data Source
AI summary
Disclosed techniques for graph generation for online path planning offer multiple advantages, such as providing for high-quality motion during online operation of the robot, while reducing the computational burden of graph generation. Achieving these competing goals involves reducing the dimensionality of the graph generation problem by performing unconstrained sampling that defines partial robot poses that set values for fewer than all configuration parameters of the robot. The remaining configuration parameters for each sample are then determined in dependence on a distance function that relates the partial pose to one or more reference robot poses that are associated with one or more tasks to be performed by the robot and are provided as inputs to the graph generation. Reference robot poses may be determined automatically based on computer analysis of the robot application or may be user-input values.