Hierarchical Deep Neural Network Path Planning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing path planning algorithms for robots, such as sampling-based and learning-based methods, face challenges in efficiently determining optimal paths in large exploring spaces while maintaining high learning convergence and reducing hardware resource requirements.
Innovation Solution
A learning method that constructs a block space by segmenting the exploring space into unit spaces, transforming absolute coordinates into relative coordinates, and using hierarchical deep neural networks for global and local path exploration, allowing for efficient path planning without requiring numerous learning networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sampling-based algorithms are used for path planning in large exploring spaces, then the robot can explore the space, but the learning convergence is poor and hardware resource requirements increase
Solution Approach 1:
The patent segments the exploring space into multiple block spaces, each containing a subset of obstacles. This segmentation allows the learning algorithm to train on smaller, manageable subsets rather than the entire large space, improving learning convergence while reducing hardware resource requirements. The robot learns path planning in divided regions and integrates these learned paths to solve the complete path planning problem.
2Reliability
If the exploring space is divided into multiple block spaces, then learning convergence improves, but the path planning process becomes more complex
Solution Approach 1:
The patent merges the path planning results from multiple block spaces to form a complete path from the starting point to the goal point. By combining the learned paths from segmented regions, the system achieves improved learning convergence while maintaining an integrated path planning process that solves the overall navigation problem.
3Productivity
If learning-based algorithms are used instead of sampling-based algorithms, then path planning efficiency improves, but the system requires more hardware resources
Solution Approach 1:
The patent applies segmentation to reduce hardware resource requirements while maintaining learning-based algorithm efficiency. By dividing the exploring space into block spaces, the system can use lighter hardware resources for each learning task while achieving overall high path planning efficiency through the integration of multiple learned paths.
Data Source
AI summary
A learning method explores, in a block space, a global path from a sub initial point to a sub goal candidate region for movement of an agent, and limits, based on the global path, an exploring space to thereby determine a limited space in the exploring space. The method arranges a sub goal in the limited space in accordance with a position of a goal point, and transforms absolute coordinates of each of at least one obstacle and a sub goal in the limited space into corresponding relative coordinates relative to a position of an agent located in the limited space. Then, the method explores, in the limited space, a target path from the initial point to the sub goal.