Multi-Object Tracking with Sensor Prioritization for Lane-Level Traffic
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for traffic information management, particularly in autonomous and non-autonomous vehicles, face limitations in providing lane-level accuracy due to reliance on road-level traffic data, which does not account for variations within lanes, and fail to prioritize sensor data based on accuracy and proximity, leading to incomplete navigation and control strategies.
Innovation Solution
A system that utilizes edge/cloud servers to receive and prioritize sensor data from connected vehicles, creating global tracklets by associating local tracklets based on ID matching, position, and direction, and assigning priority scores based on sensor accuracy and proximity, to construct a lane-level traffic map, enabling accurate lane-level traffic information and control signals for autonomous vehicles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If road-level traffic data is used for navigation, then the system is simple to implement, but lane-level accuracy is insufficient
Solution Approach 1:
The patent segments traffic data into road-level and lane-level components. Lane-level tracklets are extracted and associated with corresponding road-level tracklets, allowing the system to maintain both simple road-level processing and precise lane-level tracking simultaneously. This segmentation enables lane-level accuracy without completely redesigning the navigation system.
Solution Approach 2:
The patent adds a lane-level dimension to the existing road-level traffic data structure. By creating tracklets at both road and lane levels and establishing associations between them, the system transitions from two-dimensional road-level tracking to three-dimensional tracking that includes lane information, achieving lane-level accuracy while building upon the existing simple road-level system.
2Reliability
If all sensor data from connected vehicles is processed equally, then data utilization is comprehensive, but data quality and reliability are reduced due to noise from inaccurate sensors
Solution Approach 1:
The patent applies local quality by assigning different weights to different sensor data sources based on their individual accuracy characteristics. Each connected vehicle's sensor data is evaluated and weighted according to its specific reliability, allowing high-quality data to have greater influence while low-quality data contributes less. This ensures data reliability without discarding valuable information from vehicles with moderate sensor accuracy.
Solution Approach 2:
The patent changes the parameter of data weighting dynamically based on sensor accuracy metrics. By adjusting the weight parameters assigned to each vehicle's sensor data according to measured accuracy levels, the system optimizes the contribution of each data source. This parameter adjustment maintains comprehensive data utilization while filtering out noise from less reliable sensors through adaptive weighting.
3Measurement precision
If multiple local tracklets are merged into global tracklets with priority scoring, then tracking accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent performs preliminary actions by pre-establishing associations between lane-level and road-level tracklets before final merging. Local tracklets are first matched with corresponding global tracklets using position and direction criteria, and priority scores are pre-calculated based on sensor accuracy and proximity. This preliminary organization reduces the computational burden of the final merging operation while maintaining high tracking accuracy.
Solution Approach 2:
The patent replaces complex mechanical merging operations with algorithmic priority-based selection. Instead of equally processing all possible tracklet combinations, the system uses priority scores derived from sensor accuracy and proximity metrics to determine which local tracklets should be merged with which global tracklets. This algorithmic substitution reduces computational complexity while preserving tracking accuracy through intelligent selection.
Data Source
AI summary
Systems and methods are provided for multi-object tracking, for example by vehicles, with data source prioritization. Some embodiments of the present disclosure are directed to multi-vehicle, multi-object tracking and prioritization based on vehicle or sensor capabilities or accuracy levels.