Vehicle Time-to-Collision Prediction Using Trajectory Intersections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for determining time to collision (TTC) in intelligent driving systems are prone to inaccuracies due to single-step predictions, especially when large time intervals are involved, and are affected by perception errors, leading to poor robustness and high resource utilization.
Innovation Solution
A method that determines collision conditions and multiple trajectory intersections between an ego vehicle and a target object using vehicle and detection data to accurately calculate TTC, improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If single-step prediction and deduction is used for TTC calculation, then the calculation process is simple, but the accuracy deteriorates when time interval is large
Solution Approach 1:
The patent divides the single-step prediction process into multiple sub-steps, where the prediction is performed in segments across different time intervals. This segmentation allows the system to maintain calculation simplicity while improving accuracy by breaking down the complex prediction into manageable parts that can be processed sequentially.
Solution Approach 2:
The patent introduces a temporal dimension by considering multiple time intervals and trajectory intersections rather than a single time step. This dimensional expansion transforms the problem from a single-point prediction to a multi-point analysis, improving accuracy without proportionally increasing complexity.
2Measurement precision
If single-step prediction with small time interval is used, then the accuracy is improved, but resource utilization deteriorates
Solution Approach 1:
The patent segments the prediction process to focus computational resources only on critical trajectory intersections and collision conditions rather than continuously calculating at every small time interval. This selective segmentation maintains accuracy at critical moments while reducing overall resource consumption.
Solution Approach 2:
The patent applies partial action by performing detailed predictions only when trajectory intersections indicate potential collision risks, rather than continuously executing full prediction algorithms at every time step. This approach maintains necessary accuracy while avoiding excessive resource usage during low-risk periods.
3Productivity
If conventional single-step prediction is used, then the calculation is fast, but robustness against perception errors deteriorates
Solution Approach 1:
The patent incorporates feedback mechanisms where multiple trajectory intersections and collision conditions are evaluated to verify prediction consistency. This feedback loop allows the system to quickly identify and correct perception errors by cross-validating results across multiple prediction paths, maintaining both speed and robustness.
Solution Approach 2:
The patent performs preliminary analysis of trajectory intersections and collision conditions before final TTC determination. This preliminary action allows the system to quickly filter out obviously safe or unsafe scenarios, reserving detailed calculation only for borderline cases where perception errors might significantly impact the result, thus maintaining speed while improving robustness.
Data Source
AI summary
Disclosed are a time to collision determining method and apparatus for a vehicle. The method includes: determining vehicle data of an ego vehicle and detection data of a target object; determining a first predicted driving trajectory of the ego vehicle and a second predicted driving trajectory of the target object based on the ego vehicle data and the detection data, respectively; determining a collision condition based on first driving data of the ego vehicle data, second driving data of the detection data, and the second predicted driving trajectory; determining a plurality of trajectory intersections based on the first predicted driving trajectory and the second predicted driving trajectory; and determining a time to collision based on the ego vehicle data, the detection data, the collision condition, and the plurality of trajectory intersections.