Personalized Automatic Driving Track Planning via Style-Based Cost Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automatic driving track planning algorithms fail to personalize driving tracks for individual drivers, leading to a poor match with different driving styles and resulting in a prolonged running-in period for automatic driving.
Innovation Solution
An automatic driving track obtaining method and apparatus that identifies a driver's unique driving style coefficient, constructs a cost function based on safety, comfort, compliance, and efficiency items, and generates a personalized driving track using a three-dimensional spatial-temporal map, considering both the driver's style and surrounding vehicle dynamics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a same standard track planning algorithm is used for any driver, then the system maintains portability and basic safety, but the driving track cannot match individual driving styles resulting in poor personalization
Solution Approach 1:
The patent segments the driving style into multiple independent dimensions (steering smoothness, speed control, lane changing behavior, etc.) and represents them as separate coefficients. This allows the system to personalize tracking by combining these segmented elements rather than requiring a completely complex new system architecture.
Solution Approach 2:
The patent changes the parameters of the existing track planning algorithm by introducing driving style coefficients that modify the cost function weights. Instead of redesigning the entire system, it adjusts specific parameters (coefficients for different driving behavior aspects) to achieve personalization while maintaining the base algorithm's portability.
2Adaptability or versatility
If a standardized cost function is used for all drivers, then the system remains simple and portable, but it cannot account for individual driving styles and surrounding vehicle dynamics
Solution Approach 1:
The patent performs preliminary action by pre-defining multiple driving style templates with characteristic coefficient combinations before actual track planning. When a driver's style is identified, the corresponding pre-prepared coefficients are directly applied, avoiding information loss while keeping the system simple and portable.
3Reliability
If traditional track planning is used without driving style consideration, then the system is simple to implement, but the running-in period is prolonged as drivers must adapt to generic tracking
Solution Approach 1:
The patent implements feedback by continuously monitoring actual driver behavior during the running-in period and using this information to refine or select the most appropriate driving style coefficients. This feedback mechanism accelerates the matching process between track planning and individual driver preferences, reducing the running-in period while improving reliability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An automatic driving track obtaining method and apparatus are provided. The method includes: obtaining a driving style coefficient of a driver of a vehicle A (S101); calculating a cost function of the driver of the vehicle A based on the driving style coefficient of the driver of the vehicle A (S102), where the cost function is used to represent costs paid when the vehicle A travels from an initial node to a current node in a driving track of the vehicle A; and obtaining the driving track of the vehicle A on a first three-dimensional spatial-temporal map through calculation according to the cost function (S 103). The driving track obtained by using the method and the apparatus can match driving styles of all drivers. This improves driver satisfaction on the driving track, and lessens a running-in period in which the driver performs automatic driving.