Vehicle Route Planning Using Pareto-Seeded Multi-Objective Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional route planning algorithms for vehicles often rely on randomly generated initial routes, which can lead to inefficient optimization processes and lack of consideration for critical factors like fuel usage, travel time, and detection likelihood, especially in complex unstructured terrains.
Innovation Solution
A method that uses a first route planning algorithm to generate a set of Pareto optimal routes between defined points within a region, considering objectives like fuel consumption, travel time, and detection likelihood, and then employs a second optimization process to determine the best route from a starting to a terminal location using these selected routes as a seed set.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If randomly generated initial routes are used, then the route planning algorithm can be initiated, but the optimization process becomes inefficient and fails to consider critical factors like fuel usage, travel time, and detection likelihood
Solution Approach 1:
The patent applies preliminary action by pre-computing Pareto optimal routes between all pairs of points in the region before the actual route planning query. These pre-computed routes form a high-quality seed set that is prepared in advance, eliminating the need for random initialization during the optimization process and significantly improving both efficiency and solution quality.
Solution Approach 2:
The patent segments the route planning problem into two phases: (1) pre-computation of Pareto optimal routes between all point pairs, and (2) construction of final routes by selecting and combining segments from the pre-computed Pareto optimal routes. This segmentation allows the system to leverage pre-computed information while maintaining flexibility for specific queries.
2Reliability
If a single route planning algorithm is used, then the implementation is simple, but the solution fails to achieve Pareto optimality across multiple objectives simultaneously
Solution Approach 1:
The patent merges multiple route planning algorithms by combining a Pareto optimal route construction algorithm with a segment-based route assembly algorithm. The first algorithm generates Pareto optimal routes considering multiple objectives (fuel, time, detection), while the second algorithm efficiently assembles final routes by selecting appropriate segments from the pre-computed Pareto optimal routes, achieving both multi-objective optimality and computational efficiency.
Solution Approach 2:
The patent changes the parameter representation by using Pareto optimal routes as pre-computed segments with specific characteristics (fuel consumption, travel time, detection likelihood). These parameterized segments are stored and reused, allowing the system to efficiently query and combine routes while maintaining multi-objective optimality without re-computing from scratch.
3Speed
If Pareto optimal routes are pre-computed for all point pairs, then the route determination speed improves, but the memory requirements increase
Solution Approach 1:
The patent extracts only the essential information from pre-computed Pareto optimal routes by storing them as segmented paths with key parameters (fuel, time, detection likelihood) rather than storing complete detailed route geometries. This extraction reduces memory requirements while preserving the ability to quickly assemble optimal routes by combining selected segments.
Solution Approach 2:
The patent applies local quality by storing pre-computed Pareto optimal routes in a segmented format where each segment contains locally optimized information for specific point pairs. This allows the system to load and use only the relevant segments needed for a particular query rather than storing or processing all possible complete routes, reducing memory usage while maintaining fast route determination.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Methods and apparatus for determining a route for a vehicle (2), the method comprising: providing a set of routes, each route from a starting location to a terminal location and determined using a first algorithm; acquiring, by a processor (6), a starting location (A); acquiring,by the processor (6), a terminal location (B); using the starting and terminal locations, selecting, by the processor (6), one or more routes from the provided set of routes dependent upon the starting location (A) and the terminal location (B); and, using the acquired starting and terminal locations, and the selected routes, performing, by the processor (6), a second algorithm to determine a route for the vehicle (2) from the starting location (A) to the terminal location (B), the second algorithm comprising an optimisation process (with respect to certain objectives) using the selected routes as at least part of an initiator set of routes.