Autonomous Vehicle Intersection Path Prediction Using Dynamics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional autonomous driving systems fail to accurately respond to vehicles interacting with intersections, often generating false warnings due to neglecting connection relationships and inaccurately predicting paths, especially at high-curvature intersections where vehicles deviate from guide lines.
Innovation Solution
An apparatus for vehicle control that selects objects intersecting the vehicle's path using dynamics information and learning models to determine the most suitable exit point and predicted path, ensuring safe distance and smooth curve alignment, thereby addressing the inaccuracies in path prediction and connection relationship management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional autonomous driving systems generate predicted paths based on guide lines at intersections, then path prediction is simplified, but prediction accuracy deteriorates because vehicles often deviate from high-curvature guide lines
Solution Approach 1:
The system dynamically adapts the predicted path based on the actual driving behavior of objects. Instead of rigidly following guide lines, the system learns from observed deviations and adjusts predictions to match real-world driving patterns, allowing the path model to evolve from static guide-line-based predictions to dynamic behavior-based predictions
Solution Approach 2:
The system changes the parameters used for path prediction from purely geometric guide line parameters to include dynamic parameters such as object velocity, acceleration, and historical trajectory data. This allows the prediction model to account for vehicle dynamics and behavioral patterns rather than relying solely on static road markings
2Measurement precision
If autonomous driving systems consider connection relationships with intersections, then path prediction improves for vehicles following guidelines, but false warnings increase for vehicles ignoring connection relationships
Solution Approach 1:
The system applies partial consideration of connection relationships by evaluating multiple possible paths with varying degrees of adherence to guide lines. Instead of requiring strict compliance with connection relationships, the system generates predictions for both guideline-following paths and deviation paths, reducing false warnings by accommodating partial deviations
Solution Approach 2:
The system incorporates feedback from actual object trajectories to validate and adjust predicted paths. By comparing predicted paths with actual observed paths, the system can identify when objects are legitimately deviating from guide lines versus when there are prediction errors, thereby reducing false warnings through continuous validation
3Adaptability or versatility
If multiple candidate paths are generated for objects at intersections, then path prediction coverage improves, but computational complexity increases
Solution Approach 1:
The system segments the path prediction process into distinct stages: generating multiple candidate paths based on guide lines, evaluating each candidate against object dynamics, and selecting the most probable path. This segmentation allows the system to handle multiple paths systematically without overwhelming computational burden at any single stage
Data Source
AI summary
An apparatus for controlling a vehicle includes an object selection device configured to select an object intersecting the vehicle at an intersection existing on a driving path of the vehicle, a risk determination device configured to determine a risk during driving of the vehicle based on a predicted path of the object, and a driving control device configured to determine a driving method of the vehicle based on a risk determination result.