Transit Trip Generation Using Three-Leg Cost Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Public transportation journey planning systems often fail to provide optimal departure and arrival times, as they may include earlier or later connections that do not change the overall arrival time, leading to suboptimal routes with unnecessary delays or missed opportunities for earlier departures.
Innovation Solution
A method involving three least cost searches of a transit graph to determine optimal departure times from a source station and intermediate stations while maintaining a lowest cost arrival time at a destination station, using algorithms like Dijkstra's to compute the best, latest, and earliest possible departure and arrival times for each leg of the journey.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If a single shortest path algorithm is used to compute the best trip, then the earliest arrival time at the destination is achieved, but the system fails to identify optimal departure times from intermediate stations and may include suboptimal connections
Solution Approach 1:
The patent divides the journey computation into three separate search operations: a forward search from the source station, a backward search from the destination station, and an intermediate search. Each search computes optimal times for specific segments of the journey, allowing precise identification of optimal departure times at each station rather than relying on a single aggregate path finding algorithm.
Solution Approach 2:
The patent employs a backward search that traverses the transit graph in reverse, starting from the destination station and computing latest departure times that still allow arrival by the optimal time. This inverted approach complements the forward search by providing constraints from the destination perspective, enabling identification of optimal transfer windows at intermediate stations.
2Productivity
If pre-computed transfer patterns are used to reduce data processing, then query speed is improved, but the system may return different results with the same cost without identifying truly optimal routes
Solution Approach 1:
The patent performs preliminary computation of optimal times at each station through forward and backward searches before the final route assembly. By pre-computing earliest arrival times from the source and latest departure times to the destination for each intermediate station, the system establishes precise time windows for optimal transfers, ensuring that subsequent route selection is based on verified optimal timing rather than approximate cost calculations.
3Adaptability or versatility
If the system accepts multiple results with the same cost, then more route options are provided, but optimal departure times from source and intermediate stations are not identified
Solution Approach 1:
The patent applies local quality optimization by computing precise optimal times for each specific station along the route rather than applying a uniform cost-based selection across all routes. The forward search determines the earliest arrival time at each intermediate station, while the backward search determines the latest departure time from each intermediate station that still allows arrival by the optimal time. This station-specific time optimization ensures that each segment of the journey is timed optimally.
Data Source
AI summary
Systems and methods for generating transit trips between an origin and a destination are provided. Searches can be undertaken to identify optimal departure times from a source station and/or one or more intermediate stations while maintaining a lowest cost arrival time at a destination station. In this manner, public transportation journey schedules for routes can be determined and recommendations can be provided even if different journey schedules show identical costs.