Point Confidence Scores for Obstacle Trajectory Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional motion planning in autonomous vehicles does not accurately account for differences in vehicle types, leading to potential inaccuracies and inefficiencies in navigating around moving obstacles like other vehicles, cyclists, and pedestrians, as it relies on curvature and speed without considering specific features of various vehicles.
Innovation Solution
A process that generates multiple confidence scores along a predicted trajectory of a moving obstacle, based on its current status, map data, and traffic rules, allowing the autonomous vehicle to adjust its route accordingly to avoid potential collisions by determining point confidence scores influenced by environmental factors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional motion planning estimates path difficulty based only on curvature and speed, then the planning process is simple and fast, but the accuracy and adaptability to different vehicle types deteriorate
Solution Approach 1:
The patent segments the trajectory prediction into multiple discrete points along the predicted path. Each point receives an individual confidence score based on local environmental factors, rather than treating the entire trajectory as a single entity. This segmentation allows for more nuanced accuracy assessment while maintaining computational efficiency.
Solution Approach 2:
The patent applies local quality by assigning different confidence levels to different segments of the trajectory based on local environmental conditions. Each trajectory point's confidence score is determined by specific environmental factors at that location, such as proximity to obstacles, road geometry, and traffic conditions, rather than applying a uniform confidence level to the entire path.
2Adaptability or versatility
If a single confidence score is assigned to the entire trajectory, then the system is simple to implement, but it cannot provide nuanced guidance for route adjustments at specific points
Solution Approach 1:
The trajectory confidence scoring system is segmented into multiple point-specific confidence scores along the predicted path. Each segment or point on the trajectory receives its own confidence evaluation based on local environmental factors, enabling flexible and nuanced route adjustment decisions at specific locations rather than treating the entire trajectory uniformly.
Solution Approach 2:
The confidence scoring system is dynamic and adaptive, adjusting confidence levels at different trajectory points based on real-time environmental conditions. The system can dynamically modify route planning decisions by identifying specific high-risk segments with low confidence scores and adjusting the vehicle path accordingly, rather than using a static single-score approach.
3Reliability
If the autonomous vehicle closely follows the predicted obstacle trajectory to avoid collisions, then collision avoidance improves, but the vehicle may make excessive route adjustments reducing travel efficiency
Solution Approach 1:
The patent applies local quality by making route adjustments only at specific trajectory points where confidence scores indicate elevated risk, rather than continuously deviating from the predicted obstacle path. This localized approach maintains collision avoidance reliability by focusing safety measures on high-risk segments while preserving travel efficiency by following the original trajectory through low-risk areas.
Solution Approach 2:
The system applies partial action by selectively adjusting the route only where necessary based on point-specific confidence scores. Rather than making comprehensive route adjustments along the entire predicted trajectory, the vehicle makes targeted modifications only at points where environmental factors suggest potential hazards, thereby maintaining overall travel efficiency while ensuring safety where needed.
Data Source
AI summary
In one embodiment, a process is performed during controlling Autonomous Driving Vehicle (ADV). A plurality of point confidence scores are determined, each defining a reliability of a corresponding point on a trajectory of a moving obstacle. At least one of the point confidence scores is determined based on a) an overall trajectory confidence score, and b) at least one environmental factor of the obstacle. The ADV is controlled based on the trajectory of the moving obstacle and at least one of the plurality of point confidence scores.