Lane Network Graph for Vehicle Tracking Stability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Two-dimensional tracking systems face instability and tuning challenges due to arbitrary target motion, making it difficult to model and implement effectively, especially in automotive applications where sensors provide asynchronous data with varying characteristics.
Innovation Solution
A lane network graph is used to restrain target motion, integrating data from lidar and radar sensors by projecting observations onto a graph network, leveraging sensor characteristics to filter out noise and estimate target position using a Kalman filter, while synchronizing observations and correcting for time delays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a two-dimensional tracking system is used to track targets in open space, then the system can detect targets at any position, but the system becomes unstable and difficult to tune due to arbitrary motion patterns
Solution Approach 1:
The patent segments the two-dimensional tracking space into one-dimensional lane segments based on road network graphs. Instead of tracking targets in continuous 2D space, the system divides the space into discrete lane segments that targets must follow, making the tracking problem more manageable and stable while preserving the ability to detect targets at various positions along these segments
Solution Approach 2:
The patent introduces a road network graph as an intermediary structure between the sensors and the tracking algorithm. This graph serves as a mediator that constrains and organizes target positions along realistic vehicle paths, transforming the arbitrary 2D motion problem into a structured 1D motion problem along graph edges while maintaining versatility in detecting targets at different locations
2Adaptability or versatility
If data from multiple sensors with different modalities is integrated, then the tracking system can leverage diverse sensor characteristics, but the system becomes complex in handling asynchronous data with varying time delays and accuracy
Solution Approach 1:
The patent performs preliminary actions by predicting target positions at standardized timestamps before integrating sensor data. The system uses motion models to predict where targets should be at specific time points, then compares these predictions with actual sensor observations. This preliminary prediction step simplifies the integration of asynchronous data from multiple sensors by providing a common reference framework
Solution Approach 2:
The patent changes the temporal parameters of sensor data by resampling asynchronous observations to synchronized timestamps. Instead of dealing with the original asynchronous timing of each sensor, the system transforms all sensor inputs to a common time reference frame using prediction and interpolation, thereby simplifying the integration process while preserving the complementary information from different sensor modalities
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach results in a more efficient and robust tracking system that correctly integrates sensor data, reduces false targets, and enables accurate obstacle detection and collision avoidance in autonomous vehicles.
Implementation Method 1
The first sensor can be a lidar sensor
Implementation Method 2
The second sensor can be a radar sensor
Data Source
AI summary
A vehicle system includes a first sensor and a second sensor, each having, respectively, different first and second modalities. A controller includes a processor configured to: receive a first sensor input from the first sensor and a second sensor input from the second sensor; detect, synchronously, first and second observations from, respectively, the first and second sensor inputs; project the detected first and second observations onto a graph network; associate the first and second observations with a target on the graph network, the target having a trajectory on the graph network; select either the first or the second observation as a best observation based on characteristics of the first and second sensors; and estimate a current position of the target by performing a prediction based on the best observation and a current timestamp.