Uncertain Road User Trajectory Prediction via Drivable-Area Exit Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting the trajectories of uncertain road users (URUs) in autonomous vehicles are inadequate, particularly in urban environments, as they often rely on single entry or exit point assumptions, leading to false classifications and inefficient collision avoidance strategies.
Innovation Solution
The system identifies multiple potential entry and exit points for URUs within drivable areas using perception information and a scoring mechanism based on loss functions, allowing for more accurate trajectory prediction and collision avoidance by analyzing real-time URU and environmental data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple goal points are identified and scored using perception information and loss functions, then trajectory prediction accuracy is improved, but device complexity increases
Solution Approach 1:
The drivable area is segmented into multiple discrete goal points rather than treating it as a continuous space. This segmentation allows the system to evaluate specific locations independently, improving prediction accuracy by considering multiple potential exit points while managing complexity through structured discretization of the prediction space.
Solution Approach 2:
The system performs preliminary identification and scoring of multiple goal points before final trajectory prediction. By pre-computing scores for various exit points based on perception information and loss functions, the system prepares prediction candidates in advance, which improves accuracy while organizing complexity into manageable preprocessing and prediction stages.
2Reliability
If multiple potential entry and exit points are analyzed, then collision avoidance effectiveness is improved, but computational time increases
Solution Approach 1:
The system evaluates multiple goal points beyond what a single-point approach would provide, but not exhaustively all possible points. By selecting a reasonable number of potential entry and exit points to analyze, the system achieves improved collision avoidance effectiveness while limiting computational time through controlled evaluation scope rather than exhaustive analysis.
Solution Approach 2:
The system uses loss functions that incorporate feedback from perception information to score and rank goal points. This feedback mechanism allows the system to efficiently prioritize the most likely exit points based on real-time data, improving collision avoidance by focusing computational resources on high-probability scenarios while reducing overall computational time through intelligent filtering.
Data Source
Figure 1
Figure 2A~2B
Figure 3
AI summary
Methods and systems for controlling navigation of a vehicle are disclosed. The system will first identify a plurality of goal points corresponding to a drivable area that a vehicle is traversing or will traverse, where the plurality of goal points are potential targets that an uncertain road user (URU) within the drivable area can use to exit the drivable area. The system will then receive perception information relating to the URU within the drivable area, and identify a target exit point from the plurality goal points based on a score. The score is computed based on the received perception information and a loss function. The system will generate a trajectory of the URU from a current position of the URU to the target exit point, and control navigation of the vehicle to avoid collision with the URU.