Lane-Level Speed Profiles Using K-Means Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital maps rely on crowd-sourced probe data for vehicle speed profiles, which are computationally intensive and limited by processing power, leading to less accurate lane-level speed information for route optimization and autonomous vehicle control.
Innovation Solution
Implementing a k-means clustering algorithm to aggregate and cluster historical vehicle speed data at a lane level, generating lane-level speed profiles for road segments and strands, allowing for more precise navigation and autonomous vehicle control by identifying optimal travel lanes based on desired speeds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If crowd-sourced probe data is used for vehicle speed profiles, then route optimization can be provided, but the computational intensity is high and processing power is limited
Solution Approach 1:
The system performs preliminary actions by pre-computing and storing speed profiles for different lanes, road segments, and time epochs before they are needed for route optimization. Historical probe data is aggregated and processed in advance to create lookup tables of speed profiles, which can then be quickly retrieved during real-time navigation without intensive computation
Solution Approach 2:
The system segments the road network into discrete road segments and further segments speed data by lane, time epoch, and traffic conditions. This segmentation allows the computational workload to be divided into manageable pieces that can be pre-processed and stored, reducing the computational intensity required during real-time route optimization
2Productivity
If aggregate traffic data is used, then route guidance can be provided, but lane-level speed information accuracy is reduced
Solution Approach 1:
The system applies local quality by maintaining different levels of data granularity for different purposes. At the lane level, it stores detailed speed profiles for precise lane selection, while at the road segment level, it maintains aggregate data for overall route planning. This allows the system to provide accurate lane-level guidance when needed without the computational burden of processing all detailed data continuously
3Measurement precision
If detailed lane-level speed profiles are generated, then navigation precision is improved, but data processing complexity increases
Solution Approach 1:
The system performs preliminary actions by pre-computing lane-level speed profiles from historical probe data and storing them in an organized structure indexed by road segment, lane, and time epoch. This preliminary processing transforms complex raw data into ready-to-use speed profiles that can be quickly retrieved during navigation without real-time computation
Solution Approach 2:
The system adds temporal dimension by organizing speed profiles into different time epochs (e.g., rush hour, off-peak, nighttime). This dimensional organization allows the system to provide precise lane-level guidance by selecting appropriate profiles based on current time conditions, reducing the need to process all historical data during navigation
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method is provided for establishing lane-level speed profiles for road segments and strands of road segments based on historical vehicle probe data. Methods may include: receiving a plurality of probe data points, where a sequence of probe data points from a respective probe apparatus defines a trajectory of the respective probe apparatus; map-matching trajectories of the plurality of probe apparatuses to lanes of a road segment of a road network; determining, from each trajectory of the plurality of probe apparatuses to lanes of a road segment of a road network; determining, from each trajectory of the plurality of probe apparatuses map-matched to the lanes of the road segment, average path speeds along each lane of the road segment; clustering the aggregated speeds on each lane into a predetermined number of speed profiles using a clustering technique; and generating a lane-level speed profile for the road segment.