GPU-Accelerated Route Planning with Overlay Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current road-mapping technologies are inefficient in computing shortest paths due to high computational time and resource requirements, especially when handling various metrics such as traffic information and personal preferences, and are not adequately equipped to handle real-time updates and customizations.
Innovation Solution
A point-to-point shortest path technique that supports real-time queries and fast metric updates, utilizing a three-stage process: preprocessing, metric customization, and query stages, with preprocessing based on graph structure, metric customization performed on a GPU, and query stage using bidirectional Dijkstra's algorithm on an overlay graph and original graph components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional road-mapping programs use Dijkstra's method or heuristic variations to compute shortest paths, then route calculation can be performed, but computational time and resources become excessively large
Solution Approach 1:
The patent divides the graph into loosely connected components (cells) of bounded size and creates an overlay graph by replacing each component with a clique connecting boundary vertices. This segmentation reduces the graph size for query processing while preserving essential connectivity information, enabling faster shortest path computation.
Solution Approach 2:
The patent performs preprocessing offline to partition the graph into cells and build the overlay graph structure. This preliminary action prepares the data in advance so that during query time, only the relevant portions need to be processed, significantly reducing computation time for individual route calculations.
2Adaptability or versatility
If preprocessing is rerun for each new metric (traffic information, personal preferences, etc.), then metric customization is achieved, but computational cost increases and query times become non-competitive
Solution Approach 1:
The patent separates the graph structure (topology) from the metric (cost function). The graph is partitioned into cells with overlay graphs that are metric-independent. Only the edge costs need to be updated for different metrics, not the entire graph structure, enabling fast metric customization without rerunning preprocessing.
Solution Approach 2:
The patent extracts the metric-specific data (edge costs) from the graph structure. The overlay graph and cell partitions are maintained as permanent structures, while only the edge cost attributes are updated when new metrics are introduced, significantly reducing the computational burden of metric customization.
3Productivity
If the graph is partitioned into cells and overlay graphs are created, then query efficiency is improved, but device complexity increases
Solution Approach 1:
The overlay graph structure serves multiple purposes: it enables fast shortest path queries, supports metric customization by separating topology from costs, and allows efficient graph partitioning. This multi-functionality justifies the added structural complexity by delivering benefits across multiple operational aspects.
Data Source
AI summary
Customizable route planning is a technique for computing point-to-point shortest paths in road networks. It includes three phases: preprocessing, metric customization, and queries. A graphics processing unit may be used, e.g., in the metric customization phase, to make customization even faster, enabling a wide range of applications including highly dynamic applications and on-line personalized cost functions.