Vehicle Collision Prediction Using Dynamic Trajectory Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current driver assistance systems are limited in accurately predicting collisions across a wide range of scenarios, leading to either inaccurate activation of collision-mitigation functions or false interventions, as they struggle to model dynamic interactions between vehicles and objects with varying speeds, accelerations, headings, and curvatures.
Innovation Solution
A computer-implemented method that acquires information about moving objects, calculates multiple possible vehicle and object trajectories within confidence intervals, and performs collision calculations to determine if objects will be closer than a predetermined distance, initiating mitigation processes such as warnings or braking only when a predetermined number of calculations indicate a collision is likely.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a forward collision-mitigation function is used to detect objects and calculate time to collision, then collision prediction is simple and fast, but it cannot accurately model collisions where both the host vehicle and the object have dynamic speeds, accelerations, headings and curvatures
Solution Approach 1:
The system dynamically adapts the collision prediction model based on the detected motion characteristics of objects. When objects exhibit dynamic behavior (varying speeds, accelerations, headings, curvatures), the system switches from a simple forward collision-mitigation model to a more complex dynamic trajectory model that accounts for these variations, thereby maintaining accuracy across diverse collision scenarios.
Solution Approach 2:
The system changes the parameters used in collision calculation based on the scenario. For simple scenarios, it uses basic time-to-collision parameters. For complex scenarios involving dynamic objects, it incorporates additional parameters such as acceleration, heading rate, and curvature to accurately predict collision risk, thus resolving the contradiction between simplicity and comprehensiveness.
2Adaptability or versatility
If a driver assistance system monitors a wide range of collision scenarios, then it can predict more types of collisions, but it becomes prone to false intervention due to the broader range of objects monitored
Solution Approach 1:
The system applies partial monitoring intensity to different object categories. Instead of uniformly monitoring all objects with the same level of scrutiny, it adjusts the monitoring depth based on object type and risk level. This selective approach allows comprehensive scenario coverage while reducing false interventions by focusing computational resources on high-risk situations.
Solution Approach 2:
The system replaces simple geometric collision detection with more sophisticated probabilistic models that incorporate object behavior patterns, confidence intervals, and trajectory uncertainty. This substitution enables the system to distinguish between genuine collision risks and false positives, maintaining high versatility while improving reliability.
3Measurement precision
If multiple possible trajectories are calculated for vehicle and objects within confidence intervals, then collision prediction becomes more accurate, but the computational complexity increases
Solution Approach 1:
The system segments the trajectory calculation into discrete confidence interval levels (e.g., 68%, 95%, 99% confidence). Instead of calculating all possible trajectories continuously, it divides the prediction space into manageable segments and performs collision checks at key intervals, thereby maintaining precision while reducing computational burden through structured discretization.
Solution Approach 2:
The system performs preliminary filtering of trajectories before detailed collision analysis. It first identifies promising trajectory combinations based on basic geometric constraints and motion trends, then applies full collision calculations only to these pre-selected cases. This preliminary action reduces the number of computationally intensive operations while preserving prediction accuracy.
Data Source
AI summary
Described are systems and methods for predicting road collisions between a host vehicle and one or more objects in the surroundings of the vehicle. In aspects, range rate and heading information is acquired regarding a moving object in the surroundings of the vehicle; possible vehicle trajectories of the vehicle are calculated; possible object trajectories are calculated from the range rate of the object, the heading of the object, and a possible turn and acceleration of the object; and a collision calculation is performed.