Multi-Corridor Lane Change Planning for Autonomous Vehicle Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for autonomous vehicles to determine optimal lane change locations are inaccurate due to insufficient real-time traffic and environmental context, leading to potential route failures and inefficiencies, as they rely on discrete lane graph representations that do not account for continuous trajectory planning and perception data.
Innovation Solution
The method involves generating a multi-corridor representation of local route regions, computing static and dynamic costs for lane segments, and dynamically deciding on lane changes using perception data, allowing for continuous trajectory planning and improved route adaptability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a discrete lane graph representation is used for route planning, then the routing process is simplified and computationally efficient, but the accuracy of lane change location determination deteriorates and perception data cannot be properly integrated
Solution Approach 1:
The patent segments the continuous road space into multiple corridors, where each corridor represents a sequence of lane segments. This segmentation allows the system to maintain computational efficiency while providing sufficient granularity to integrate perception data and determine accurate lane change locations within each corridor context.
Solution Approach 2:
The patent adds a new dimension to the traditional lane graph by organizing lane segments into corridors that extend in the direction of travel. This dimensional transformation allows the system to simultaneously maintain routing efficiency and improve lane change location accuracy by providing contextual information about the road geometry and traffic conditions along each corridor.
2Force
If lane change locations are determined at the routing stage without real-time context, then the routing plan can be generated in advance, but the accuracy of lane change maneuver execution deteriorates due to insufficient traffic condition information
Solution Approach 1:
The system performs preliminary route planning and identifies potential lane change corridors in advance, but delays the final lane change execution decision until real-time perception data is available. This allows the system to prepare the routing plan ahead of time while ensuring that actual lane change maneuvers are executed only when traffic conditions confirm their feasibility.
Solution Approach 2:
The patent implements a dynamic approach where the lane change decision is not fixed at routing stage but is dynamically adjusted based on real-time perception data. The system continuously evaluates whether the pre-planned lane change locations remain valid given current traffic conditions, and can modify or cancel lane change maneuvers as needed.
3Device complexity
If the lane graph is discretized for route planning, then computational complexity is reduced, but the ability to account for continuous trajectory planning and perception data deteriorates
Solution Approach 1:
The patent segments the continuous road environment into discrete corridors composed of lane segments, creating a hierarchical representation that balances computational simplicity with continuous trajectory planning capability. Each corridor maintains the continuous geometric information needed for trajectory planning while the discrete structure enables efficient computation.
Solution Approach 2:
The corridor-based representation serves multiple functions simultaneously: it provides a discrete structure for efficient route planning, maintains continuous geometric information for trajectory planning, and integrates perception data for accurate lane change location determination. This multi-functional design eliminates the need for separate processing systems.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and a system for maneuvering an autonomous vehicle is disclosed. The system includes an autonomous vehicle including one or more sensors and a processor. The processor is configured to generate a nominal route from a start position toward a destination with reference to a road network map. The nominal route includes a plurality of consecutive lane segments from the start position to the destination. The processor is further configured to use the road network map to identify at least one candidate lane segment corresponding to one or more of the plurality of consecutive lane segments to generate an expanded route representation, generate a multi-corridor representation of a local region around the autonomous vehicle while travelling on the nominal path, and generate a trajectory for the autonomous vehicle to traverse the local region using the multi-corridor representation and perception data corresponding to the autonomous vehicle.