Lane Change Intent Prediction for Autonomous Collision Avoidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles struggle to accurately predict the intent of surrounding vehicles when they intend to change lanes, leading to potentially unsafe and unnatural driving maneuvers due to uncertainty and data noise in perception systems.
Innovation Solution
The system employs a processor to monitor the lateral position and velocity of vehicles within a predetermined distance, execute a computer model to predict collision likelihood, and determine an alternative trajectory for the autonomous vehicle to avoid potential collisions based on the predicted intent of surrounding vehicles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the autonomous vehicle uses basic perception data to determine lane change intent, then the system complexity is low, but the prediction accuracy is insufficient leading to unsafe maneuvers
Solution Approach 1:
The system performs preliminary actions by continuously monitoring lateral position and velocity data before a lane change actually occurs. The computer model executes predictive calculations in advance to identify vehicles with high probability of lane change intent, allowing the autonomous vehicle to prepare appropriate responses before the lane change happens, thereby improving prediction accuracy without excessive complexity
Solution Approach 2:
The prediction system is segmented into distinct functional components: data collection module for gathering lateral position and velocity, computer model module for executing prediction algorithms, and decision module for determining collision likelihood. This segmentation allows each component to be optimized independently, improving overall accuracy while managing system complexity through modular architecture
2Reliability
If the autonomous vehicle reacts to lane changes after they occur, then the response time is short, but the safety is compromised due to lack of predictive capability
Solution Approach 1:
The system applies preliminary action by using the computer model to predict future lane change positions and trajectories before the actual lane change occurs. This allows the autonomous vehicle to identify potential collision risks in advance and prepare evasive maneuvers, improving safety without requiring excessive reaction time when the lane change actually happens
Solution Approach 2:
The system performs preliminary anti-action by identifying vehicles with high probability of lane change intent and pre-calculating collision likelihood. This allows the autonomous vehicle to take preventive measures before the lane change occurs, such as adjusting speed or preparing to change lanes, thereby improving safety while maintaining appropriate response timing
3Measurement precision
If the autonomous vehicle monitors all surrounding vehicles continuously, then the detection coverage is complete, but the computational load increases leading to processing delays
Solution Approach 1:
The system applies local quality by focusing monitoring resources on vehicles that exhibit specific characteristics indicating high lane change probability, such as vehicles within a predetermined distance from lane lines or with abnormal lateral velocity patterns. This selective monitoring approach maintains high detection accuracy for critical vehicles while reducing computational load by not uniformly monitoring all surrounding vehicles
Solution Approach 2:
The system uses partial action by monitoring lateral position and velocity data with high precision for vehicles of interest, while using less intensive monitoring for other vehicles. The computer model executes predictions only for vehicles meeting specific criteria, providing sufficient monitoring accuracy for critical cases without the computational burden of continuous full-spectrum monitoring
Data Source
AI summary
A method comprises periodically monitoring, by a processor, lateral position and velocity of vehicle within a predetermined distance from an autonomous vehicle, the vehicle moving in a direction having at least one common attribute with the autonomous vehicle; executing, by the processor, a computer model using the monitored lateral position and velocity of the vehicle, to predict whether a trajectory for the vehicle; and when a current trajectory of the autonomous has a likelihood of collision with the predicted trajectory of the vehicle that satisfies a threshold, determining, by the processor, an alternative trajectory for the autonomous vehicle.