Autonomous Driving Behavior Prediction via Stimulus-Driven Classification
Solution Overview
Problem
Autonomous driving systems fail to incorporate a driver's logic, attentive behavior, and casual reactions, limiting their ability to adapt to dynamic driving scenarios effectively.
Innovation Solution
A computer-implemented method and system that learns naturalistic driving behavior by analyzing vehicle dynamic data and image data to detect and classify behavioral events, predicting goal-oriented actions, and building a data set for autonomous vehicle control, using a four-layer annotation scheme to understand driver intentions and reactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional sensor data processing is used in autonomous driving systems, then basic object detection and road recognition are achieved, but the system fails to capture driver's logic, attentive behavior, and casual reactions
Solution Approach 1:
The patent creates a virtual copy of human driver behavior through behavioral event detection and classification. By analyzing vehicle dynamic data and image data to detect behavioral events and classify them into stimulus-driven and goal-oriented actions, the system replicates human driving logic, attentiveness, and reactions without requiring actual human drivers in the loop.
Solution Approach 2:
The system performs preliminary classification of behavioral events into stimulus-driven and goal-oriented actions before making autonomous driving decisions. This preliminary analysis of driver behavior patterns enables the autonomous system to anticipate and adapt to dynamic driving scenarios more effectively.
2Measurement precision
If autonomous driving systems use basic sensor data processing, then real-time object detection is achieved, but higher level situational understanding is lacking
Solution Approach 1:
The patent segments driver behavior into distinct categories: behavioral events, stimulus-driven actions, and goal-oriented actions. This segmentation allows the complex task of situational understanding to be broken down into manageable classification steps, improving measurement precision without overwhelming system complexity.
Solution Approach 2:
The system adds a new dimension of analysis by classifying behavioral events based on their motivational structure (stimulus-driven vs. goal-oriented). This dimensional classification enhances situational understanding by providing deeper insight into driver intentions beyond basic object detection.
3Loss of information
If the system analyzes vehicle dynamic data and image data to detect behavioral events, then driver behavior understanding is improved, but data processing complexity increases
Solution Approach 1:
The patent extracts only the essential behavioral information from vehicle dynamic data and image data by detecting and classifying specific behavioral events. This extraction approach retains critical driver behavior information while filtering out unnecessary data, managing processing complexity effectively.
Data Source
AI summary
A system and method for learning naturalistic driving behavior based on vehicle dynamic data that include receiving vehicle dynamic data and image data and analyzing the vehicle dynamic data and the image data to detect a plurality of behavioral events. The system and method also include classifying at least one behavioral event as a stimulus-driven action and predicting at least one behavioral event as a goal-oriented action based on the stimulus-driven action. The system and method additionally include building a naturalistic driving behavior data set that includes annotations that are based on the at least one behavioral event that is classified as the stimulus-driven action. The system and method further include controlling a vehicle to be autonomously driven based on the naturalistic driving behavior data set.