Travel Plan Generation Using Recurrent Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle routing problem solutions using machine learning, such as recurrent neural networks with attention mechanisms, often struggle to find optimal travel plans due to restrictive output rules when multiple vehicles are involved, leading to suboptimal solutions.
Innovation Solution
A travel plan generation device utilizing a recurrent neural network with an attention mechanism that selects points and vehicles based on input information to generate routes dynamically, allowing for the optimization of routes by vehicles visiting multiple points without predetermined order constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If a recurrent neural network with attention mechanism is used to solve the vehicle routing problem, then the travel plan can be acquired in a short time, but the solution is not strictly optimal due to the enormous number of route patterns
Solution Approach 1:
The patent segments the vehicle routing problem into multiple sub-problems by dividing the set of visiting points into different groups, each handled by a dedicated neural network model. This segmentation allows each model to focus on a specific subset of points, making the optimization more manageable and improving solution quality while maintaining computational efficiency.
Solution Approach 2:
The patent employs dynamic neural network models that can adapt their structure and parameters based on the input problem characteristics. The models dynamically adjust their routing strategies by learning from the spatial relationships and constraints of specific problem instances, enabling them to produce near-optimal solutions faster than traditional methods.
2Ease of manufacture
If a rule is imposed that vehicles select visiting points in a predetermined order, then the neural network can output a travel plan, but the output is restricted and may not be optimal
Solution Approach 1:
Instead of imposing a predetermined order on vehicle visits, the patent inverts the approach by allowing the neural network to learn the optimal visiting sequence from the data. The model processes point information and dynamically determines the best order to visit points based on spatial relationships and constraints, rather than following a fixed predetermined sequence.
Solution Approach 2:
The patent changes the parameter of visiting order from a fixed predetermined sequence to a dynamic learned sequence. The neural network learns optimal visiting orders by processing spatial coordinates and constraints, allowing the system to adapt the visiting sequence to each specific problem instance rather than using a universal predetermined order.
3Adaptability or versatility
If the number of vehicles and visiting points is large, then the problem covers more real-world scenarios, but the number of route patterns becomes enormously large making it difficult to acquire optimal solutions
Solution Approach 1:
The patent divides the large set of visiting points into multiple subsets, with each subset handled by a separate neural network model. This segmentation reduces the effective problem size for each model, making it computationally feasible to handle large real-world scenarios while maintaining solution quality. The segmented approach manages the combinatorial complexity by breaking it into smaller, more tractable sub-problems.
Data Source
AI summary
A travel plan generation device according to an aspect of the present invention is provided with a generation unit that generates a travel plan for traveling a plurality of points by a plurality of mobile bodies by performing, at each output step, processing of selecting any one point out of the plurality of points and any one mobile body out of the plurality of mobile bodies by using a recurrent neural network configured to output visiting probabilities at the plurality of points and use probabilities of the plurality of mobile bodies when point information regarding the plurality of points and mobile body information regarding the plurality of mobile bodies are input, and an output unit that outputs the travel plan.