Autonomous Vehicle Trajectory Clustering for Reactive Path Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in efficiently navigating through complex environments due to high computational load and time requirements, often resulting in stuttering or the need for remote operator intervention, especially in scenarios with dynamic objects like traffic cones or pedestrians.
Innovation Solution
An autonomous vehicle guidance system that uses a machine-learned model to cluster candidate trajectories based on static and dynamic object maps, reducing the number of predictions needed by classifying objects as reactive or passive and determining a representative trajectory to reduce computational load and improve navigation efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the vehicle evaluates all candidate trajectories to ensure safe navigation, then navigation safety is improved, but computational time and processing load increase significantly
Solution Approach 1:
The patent segments the evaluation process by dividing candidate trajectories into clusters based on similarity metrics. Instead of evaluating each trajectory individually, the system identifies representative trajectories from each cluster and evaluates only those representatives, significantly reducing the number of evaluations required while maintaining comprehensive coverage of the trajectory space.
Solution Approach 2:
The patent performs preliminary clustering and identification of representative trajectories before the actual safety evaluation. This preliminary action organizes the trajectory space in advance, allowing the evaluation phase to proceed efficiently with a reduced set of candidate trajectories that already represent diverse navigation options.
2Object-affected harmful factors
If the vehicle evaluates every candidate trajectory to avoid unsafe paths, then collision avoidance is improved, but computational load becomes excessive
Solution Approach 1:
The patent merges multiple similar trajectories into clusters based on their geometric and kinematic properties. By combining trajectories that share similar characteristics, the system reduces the total number of independent evaluations needed while ensuring that the representative trajectory from each cluster captures the safety characteristics of all merged trajectories.
Solution Approach 2:
The patent uses representative trajectories as copies that stand in for multiple similar trajectories. Each representative trajectory is selected to embody the essential safety characteristics of its cluster members, allowing the system to evaluate one copy instead of multiple originals while maintaining equivalent safety assurance.
3Measurement precision
If the vehicle uses complex trajectory evaluation to handle dynamic objects accurately, then navigation accuracy is improved, but processing speed decreases
Solution Approach 1:
The patent implements a dynamic evaluation strategy where the level of detail in trajectory evaluation adapts based on the characteristics of the trajectory cluster. For clusters with high internal similarity, less computationally intensive evaluation methods are used, while clusters with greater diversity receive more thorough evaluation, optimizing the balance between accuracy and speed.
Data Source
AI summary
The techniques discussed herein may comprise an autonomous vehicle guidance system that generates a group of candidate trajectories based at least in part on sensor data. This group of candidate trajectories is clustered into two or more clusters and one or more representative trajectories may be determined for each cluster. The representative trajectory may be one of the trajectories in the cluster or a separately determined trajectory that represents the trajectories in a cluster, such as a mean or median trajectory. The representative trajectory may be used to predict how a dynamic object would react to the representative trajectory. This prediction may be used to determine potentially different costs associated with the different trajectories of the cluster with which the representative trajectory is associated. These costs may include costs to induce following behavior by the autonomous vehicle and/or cost(s) associated with conditionally available roadway portions.