Lane-Level Route Planner Explainability for Slack-Based Decisions
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 provide precise lane changes and contingency planning, and are not explainable, which is crucial for ensuring safety and user trust.
Innovation Solution
A lane-level route planning method that incorporates lane-specific information, learns from historical data, and provides explainable decisions by using a navigation map that includes probabilities and objectives, allowing for multi-objective optimization and real-time adjustments, enabling autonomous vehicles to make informed lane changes and handle contingencies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional road-level route planning is used, then the route planning system is simple to implement, but it lacks lane-level information required for autonomous driving
Solution Approach 1:
The patent segments the route planning problem into two distinct levels: road-level planning (for overall route determination) and lane-level planning (for specific lane selection and maneuver execution). This segmentation allows the system to maintain simplicity at the road level while adding detailed lane-level information only where necessary for autonomous driving, thus improving reliability without proportionally increasing overall system complexity.
Solution Approach 2:
The patent adds a new dimension of abstraction by introducing lane-level planning as a separate layer above the traditional road-level planning. Instead of merely enhancing the existing road-level system, it creates a hierarchical structure with road-level and lane-level components, allowing the system to handle autonomous driving requirements without completely redesigning the fundamental route planning architecture.
2Measurement precision
If lane-level route planning is implemented, then precise lane changes and contingency planning are enabled, but the system complexity increases
Solution Approach 1:
The patent separates lane-level planning into distinct functional modules including lane segment identification, lane change maneuver planning, and contingency planning. Each module handles a specific aspect of lane-level decision-making, which improves measurement precision for individual functions while managing overall system complexity through modular design that allows independent development and testing of each component.
3Adaptability or versatility
If multi-objective optimization is used, then the route planner can balance time, comfort, and safety, but the computational complexity increases
Solution Approach 1:
The patent implements dynamic objective weighting where the relative importance of different objectives (time, comfort, safety) can change based on the current driving context, vehicle state, and environmental conditions. This dynamic approach allows the system to adapt to different situations without requiring a complete recomputation of all possible routes, thus improving versatility while managing computational complexity through context-aware prioritization.
Solution Approach 2:
The patent performs preliminary filtering of candidate routes and lane segments based on basic constraints before applying full multi-objective optimization. By pre-processing the search space to eliminate obviously suboptimal options, the system reduces the computational burden of the subsequent multi-objective optimization while still maintaining the ability to balance multiple objectives effectively.
4Loss of information
If the system provides detailed explanations for lane selection, then user trust and understandability improve, but information processing requirements increase
Solution Approach 1:
The patent extracts and presents only the most relevant factors influencing lane selection to the user, rather than providing a complete dump of all computational data. The explanation module identifies key decision factors (such as upcoming turns, traffic conditions, or safety considerations) and presents them in a simplified, human-understandable format, thus improving explainability while managing information processing requirements by filtering out unnecessary details.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Providing explanations in route planning includes determining a route based on at least two objectives received from a user, where a second objective of the at least two objectives is constrained to within a slack value of a first objective of the at least two objectives; receiving,from the user, a request for an explanation as to an action along the route; and providing the explanation to the user. The explanation describes an extent of violating the slack value.