Vehicle Merge Detection Using Dynamic Trajectory Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous driving systems struggle to accurately detect and manage the insertion of a target vehicle into a traffic lane, especially when navigating bends or lane changes, due to uncertainties in lane recognition and vehicle trajectory prediction, leading to incorrect vehicle selection or deselection.
Innovation Solution
A method for detecting a target vehicle using sensors to determine relative positioning and set selection and deselection thresholds based on longitudinal speed and time horizon, allowing adaptive steering adjustments without predicting lane trajectories, ensuring robust detection in straight lines and bends.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If lane recognition and trajectory prediction are used to detect target vehicles, then detection accuracy in straight lines is improved, but reliability deteriorates when navigating bends or during lane changes due to inconsistent trajectory determination
Solution Approach 1:
The patent changes the detection parameters from static lane-based coordinates to dynamic relative position parameters. The system uses longitudinal distance (along the ego vehicle's trajectory) and lateral distance (perpendicular to the trajectory) that are continuously updated based on the ego vehicle's current speed and trajectory, rather than relying on fixed lane markings. This allows accurate detection whether the vehicle is moving straight or navigating bends.
Solution Approach 2:
The patent introduces dynamic thresholds for target vehicle detection that adapt to the ego vehicle's longitudinal speed and time horizon. The selection threshold and deselection threshold are calculated based on current vehicle parameters, allowing the detection system to automatically adjust its sensitivity and accuracy requirements according to the driving situation, maintaining reliability across varying conditions.
2Measurement precision
If lane marking recognition is used to determine ego vehicle trajectory, then trajectory accuracy is improved in readable conditions, but detection robustness deteriorates when markings are masked, erased, or unreadable
Solution Approach 1:
The patent introduces an intermediary reference system based on the ego vehicle's own trajectory and position rather than external lane markings. By using the vehicle's self-determined trajectory (from sensors and navigation) as the reference frame, the system eliminates dependence on potentially unavailable lane markings while maintaining accurate relative position calculations for target vehicle detection.
3Device complexity
If fixed detection thresholds are used for target vehicle selection, then device complexity is reduced, but adaptability deteriorates when ego vehicle speed or driving conditions change
Solution Approach 1:
The patent implements dynamic thresholds that are automatically calculated based on the ego vehicle's longitudinal speed and a predetermined time horizon. The selection threshold and deselection threshold adapt in real-time to changing driving conditions, ensuring the detection system remains appropriately sensitive whether the vehicle is moving slowly or quickly, without requiring manual reconfiguration.
Solution Approach 2:
The detection system performs self-adjustment by automatically calculating appropriate thresholds based on its own measured speed and trajectory parameters. The system serves itself by using its own operational data (longitudinal speed, position) to configure its detection criteria, eliminating the need for external calibration or fixed preset values.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method and a device for detecting merging of a target vehicle into a traffic lane (103) from an autonomous vehicle, called ego vehicle (101), said method comprising the following steps: * Detecting said target vehicle; * Determining at least one item of information on the relative positioning of said target vehicle with respect to the ego vehicle (101); * Determining a trajectory of said ego vehicle (101); * Determining a minimum distance between said target vehicle and said trajectory; * Determining a selection threshold, said selection threshold representing a distance and being based on a longitudinal speed of said ego vehicle (101) and on said time horizon; * If said minimum distance is less than said selection threshold, increasing the presence indicator of said target vehicle; * Generating an autonomous driving instruction based on the presence indicator.