External Actor Trajectory Prediction for Autonomous Driving Decisions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in safely navigating through complex driving scenarios due to the inability to accurately predict the behavior of external actors, such as vehicles and pedestrians, which can lead to unsafe maneuvers and aborted goals.
Innovation Solution
A computer-implemented method that predicts external actor trajectories by receiving sensor inputs, applying object tracking, determining available goals, and comparing observed traces with expected trajectory models to estimate likelihoods, using generative models trained on real-world driving behavior and sampling algorithms to account for various parameters and scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If autonomous vehicles use basic sensor input and simple tracking methods, then the system complexity is low, but the prediction accuracy of external actor behavior is insufficient leading to unsafe maneuvers
Solution Approach 1:
The prediction system is segmented into multiple independent modules: sensor input processing, object tracking, goal determination, trajectory model generation, and likelihood comparison. Each module handles a specific aspect of behavior prediction, allowing the system to achieve high reliability through specialized sub-systems while managing overall complexity through modular architecture.
Solution Approach 2:
The system performs preliminary actions by pre-determining a set of possible goals for external actors and generating expected trajectory models for each goal before actual prediction is needed. This preparatory work allows the system to quickly compare observed traces with pre-computed models during real-time operation, improving safety response time while distributing computational complexity across different time phases.
2Measurement precision
If the system determines multiple available goals and expected trajectory models for each goal, then the prediction accuracy improves, but the computational time and processing load increase
Solution Approach 1:
The system applies partial action by determining a limited set of most relevant goals for external actors rather than exhaustively analyzing all possible goals. The goal determination module focuses on identifying plausible goals based on current context, and trajectory models are generated only for these selected goals, achieving sufficient prediction accuracy while reducing computational time and processing load.
3Reliability
If the system uses generative models trained on real-world driving behavior, then the realism and accuracy of trajectory predictions improve, but the model complexity and training requirements increase
Solution Approach 1:
The system uses copying by training generative models on recorded real-world driving behavior data to create virtual models that replicate human driving patterns. These trained models capture realistic behavior characteristics without requiring the actual physical presence of human drivers during operation. The models are copied from training data and can be deployed and reused across different autonomous vehicle systems, achieving high realism while managing model complexity through data-driven approaches.
Data Source
AI summary
A computer-implemented method of predicting an external actor trajectory comprises receiving, at a computer, sensor inputs for detecting and tracking an external actor; applying object tracking to the sensor inputs, in order track the external actor, and thereby determine an observed trace of the external actor over a time interval; determining a set of available goals for the external actor; for each of the available goals, determining an expected trajectory model; and comparing the observed trace of the external actor with the expected trajectory model for each of the available goals, to determine a likelihood of that goal.