Lane-Level Route Planning With Probabilistic State Transitions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional route planning systems lack lane-level information, making them inadequate for autonomous driving as they fail to account for specific lane changes and contingencies, relying on abstract road-level planning that does not consider the complexities required for autonomous vehicle navigation.
Innovation Solution
A method and apparatus for lane-level route planning that obtain and convert lane-level information into probabilities for a state transition function, enabling the generation of routes that include specific lane changes and contingency plans, using a navigation map that incorporates environment, vehicle, and human information to optimize multiple objectives such as time, comfort, and autonomy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If road-level route planning is used, then the route planning system is simple and computationally efficient, but it lacks lane-level information required for autonomous driving
Solution Approach 1:
The patent segments the route planning process into two distinct levels: road-level planning for high-level route determination and lane-level planning for specific lane navigation. This segmentation allows the system to maintain simplicity at the road level while incorporating detailed lane-level information when needed for autonomous driving operations.
Solution Approach 2:
The patent adds a lane-level dimension to the traditional road-level planning system. By introducing lane segments as an additional layer of abstraction between roads and vehicles, the system transitions from two-dimensional road networks to a three-dimensional planning space that includes lane-specific information, enabling autonomous vehicles to navigate with appropriate detail without completely redesigning the entire planning architecture.
2Reliability
If lane-level route planning is implemented, then lane-level information and contingency planning are available for autonomous driving, but the system complexity and computational requirements increase
Solution Approach 1:
The patent performs preliminary lane-level planning by pre-computing state transition functions and contingency routes based on historical lane-level information from multiple vehicles. This preliminary action allows the system to have lane-level plans ready in advance, reducing real-time computational requirements while maintaining high reliability for autonomous driving operations.
Solution Approach 2:
The system uses historical lane-level information collected from multiple vehicles to automatically build and update the state transition function and contingency plans without requiring manual intervention. This self-service approach allows the system to improve its lane-level planning capabilities over time while managing complexity through data-driven automation.
3Use of energy by moving object
If traditional road-level planning is used, then computational resources are conserved, but unsafe maneuvers may be required due to lack of lane-specific guidance
Solution Approach 1:
The patent applies local quality by providing detailed lane-level planning information only where and when it is needed for safe autonomous navigation, rather than uniformly across all route planning scenarios. The state transition function provides lane-specific guidance locally at critical decision points while allowing simpler road-level planning to handle less complex segments, optimizing the balance between computational energy consumption and driving safety.
Data Source
AI summary
Lane-level route planning includes obtaining lane-level information of a road, where the road includes a first lane and a second lane and the lane-level information includes first lane information related to the first lane and second lane information related to the second lane; converting the lane-level information to probabilities for a state transition function; receiving a destination; and obtaining a policy as a solution to a model that uses the state transition function.