Piecewise Semantic Trajectory Aggregation for Autonomous Vehicle Maneuver Planning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional machine learning techniques for vehicle control in autonomous vehicles are less effective in complicated traffic environments, particularly when vehicles interact in complex maneuvers, as they do not utilize piecewise semantic aggregation of trajectories for computing drivable spaces during motion planning.
Innovation Solution
A method and system that determine a drivable space trajectory for an ego vehicle by identifying a set of vehicle trajectories corresponding to a semantic driving maneuver, using a vehicle trajectory aggregation module, drivable trajectory identification module, and vehicle control selection module to perform the maneuver, incorporating piecewise semantic aggregation of trajectories for improved motion planning and vehicle control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional machine learning techniques are used for vehicle control, then the system is simple to implement, but the control effectiveness deteriorates in complicated traffic environments
Solution Approach 1:
The patent segments the drivable space computation into multiple discrete trajectory clusters, where each cluster represents a specific maneuver type (e.g., lane changing, merging). This segmentation allows the system to handle complex traffic scenarios by breaking them down into manageable, pre-computed trajectory sets, improving control effectiveness without requiring a monolithic complex system.
Solution Approach 2:
The patent performs preliminary computation of drivable spaces and trajectory clusters in advance, storing them for later use during motion planning. By pre-computing these trajectories offline or during idle periods, the system prepares ready-to-use maneuver options that can be quickly selected during real-time control, enhancing reliability without increasing real-time computational complexity.
2Measurement precision
If piecewise semantic aggregation of trajectories is used to compute drivable space, then the precision of motion planning improves, but the computational complexity increases
Solution Approach 1:
The patent applies local quality by computing drivable spaces with high precision only in locally relevant regions around the ego vehicle, rather than uniformly across the entire environment. Each trajectory cluster is computed with appropriate precision for its specific maneuver context (e.g., lane changing vs. straight driving), optimizing the balance between precision and computational complexity.
Solution Approach 2:
The patent changes parameters by representing drivable spaces as discrete trajectory clusters with specific semantic labels rather than continuous mathematical surfaces. This parameterization allows the system to achieve high precision in maneuver representation while reducing computational complexity through discrete categorization and lookup-based selection.
3Adaptability or versatility
If multiple trajectory clusters are aggregated for the same semantic maneuver, then the adaptability to different traffic scenarios improves, but the system complexity increases
Solution Approach 1:
The patent creates universal trajectory clusters that can serve multiple functions across different traffic scenarios. Each cluster represents a fundamental maneuver pattern that can be adapted to various contexts through parameter adjustment (e.g., speed, position, timing), allowing the same cluster to handle different scenarios without requiring separate specialized trajectories for each case.
Solution Approach 2:
The patent uses copying by replicating and adapting base trajectory clusters for different scenarios rather than creating entirely new trajectories each time. The system copies proven maneuver patterns and modifies them through parameter changes to fit specific contexts, reducing system complexity while maintaining high adaptability across diverse traffic situations.
Data Source
AI summary
A method of determining a drivable space trajectory of an ego vehicle is described. The method includes determining a set of vehicle trajectories corresponding to a same semantic driving maneuver during motion planning of the ego vehicle. The method also includes identifying the drivable space trajectory to perform the same semantic driving maneuver. The method further includes performing a vehicle control action to maneuver the ego vehicle along the drivable space trajectory.