Vehicle Trajectory Pairing for Short- and Long-Horizon Planning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles struggle to effectively integrate short-term and long-term trajectory planning, leading to inefficiencies and suboptimal decision-making due to the limitations of existing prediction systems.
Innovation Solution
The autonomous vehicle generates short-term and long-term trajectories in parallel, combining them to create trajectory pairings that consider both immediate and future impacts, using machine-learned models to balance granularity and foresight, thereby improving motion planning accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the autonomous vehicle generates only short-term trajectories, then the computational complexity is reduced and processing speed is improved, but the ability to account for long-term impacts and strategic goals is lost
Solution Approach 1:
The patent divides the trajectory planning into two separate segments: short-term trajectories (high granularity, immediate future) and long-term trajectories (lower granularity, extended future). This segmentation allows each segment to be processed independently with appropriate resolution, maintaining processing speed while capturing both immediate and long-term effects.
Solution Approach 2:
The patent adds a temporal dimension by introducing long-term trajectories that extend beyond the traditional short-term horizon. By incorporating trajectories with different time spans (short-term and long-term), the system gains foresight into future impacts without sacrificing the detailed short-term planning capability.
2Loss of information
If the autonomous vehicle generates both short-term and long-term trajectories, then the ability to plan with long-term foresight is improved, but the computational complexity and processing time increase
Solution Approach 1:
The patent applies different levels of detail (granularity) to different temporal regions: short-term trajectories use high granularity for immediate, critical decisions, while long-term trajectories use lower granularity for extended forecasting. This local quality differentiation reduces overall computational complexity while maintaining necessary detail where needed.
Solution Approach 2:
The patent generates a limited set of long-term trajectories (fewer than short-term trajectories) that are sufficient to capture long-term impacts without exhaustively planning every possible future path. This partial action approach provides adequate long-term foresight while avoiding excessive computational burden.
3Measurement precision
If the autonomous vehicle uses high-granularity short-term trajectories, then the precision of immediate motion planning is improved, but the ability to leverage temporal discounting for long-term uncertainty is reduced
Solution Approach 1:
The patent applies high granularity locally to short-term trajectories where precision is critical for immediate safety and control, while using lower granularity for long-term trajectories where temporal discounting is needed to manage uncertainty over extended periods. This localized approach to granularity preserves both precision and temporal discounting capability.
Data Source
AI summary
An example method includes (a) obtaining sensor data descriptive of an environment of an autonomous vehicle; (b) determining a plurality of short-term trajectories based on the sensor data; (c) determining a plurality of long-term trajectories based on the sensor data; (d) generating a first trajectory pairing based on the first short-term trajectory and the first long-term trajectory; and (e) determining, from among the plurality of short-term trajectories, a short-term trajectory for execution by the autonomous vehicle based on the first trajectory pairing.