Driver Intent Trajectory Prediction From Eye Gaze for Natural ADAS
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Advanced Driver Assistance Systems (ADAS) do not adequately consider the driver's state and intention, leading to unnatural vehicle behavior and reduced driver acceptance.
Innovation Solution
A computer-implemented method and system that determines a driver's intended trajectory by analyzing eye gazing data, using sensors and control units to adjust the vehicle's trajectory to match the driver's intended path, incorporating model predictive control and real-time adaptation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If current ADAS systems are designed without considering the driver's state and intention, then the system complexity is reduced and ease of manufacture is improved, but the driver acceptance and naturalness of vehicle behavior deteriorate
Solution Approach 1:
The system performs preliminary actions by continuously monitoring driver eye gaze direction and head position to predict the driver's intended trajectory before the driver actually executes the maneuver. This allows the ADAS to anticipate driver intentions and prepare appropriate responses, improving naturalness and acceptance without requiring complex real-time reaction systems
Solution Approach 2:
The system implements feedback by comparing the vehicle's actual trajectory with the driver's intended trajectory (derived from eye gaze and head position) and continuously adjusting the vehicle's path to align with driver intentions. This closed-loop feedback mechanism ensures the vehicle behaves naturally while maintaining safety, resolving the contradiction between system simplicity and driver acceptance
2Adaptability or versatility
If ADAS systems implement driver state monitoring and intention prediction, then driver acceptance and naturalness are improved, but the device complexity and measurement difficulty increase
Solution Approach 1:
The system achieves multi-functionality by using a single integrated monitoring setup that simultaneously captures head position, eye gaze direction, and derives intended trajectory from these measurements. This universal approach allows the system to perform multiple functions (monitoring, prediction, trajectory calculation) without proportionally increasing device complexity
Solution Approach 2:
The system introduces an intermediary computational model that translates raw sensor data (eye gaze coordinates, head position) into meaningful driver intention information. This intermediary layer simplifies the complexity by providing a standardized interface between physical sensors and control algorithms, making the overall system more manageable despite the added functionality
3Ease of operation
If the vehicle trajectory is adjusted to match driver intention in real-time, then driver comfort and acceptance are improved, but the control system complexity and processing requirements increase
Solution Approach 1:
The system applies partial action by adjusting only the lateral trajectory components that deviate from driver intention, rather than controlling all vehicle parameters. This selective approach improves driver comfort by matching natural driving behavior while avoiding the complexity of comprehensive real-time control of all vehicle systems
Solution Approach 2:
The system utilizes parameter changes by continuously updating the intended trajectory parameters based on dynamic driver eye gaze and head position measurements. This allows the control system to adapt to changing driver intentions in real-time using simple parameter adjustments rather than complex control algorithms, maintaining driver comfort while managing system complexity
Data Source
Figure 1~2
Figure 3
AI summary
A computer-implemented method to determine an intended trajectory of a driver of a vehicle is presented. The method comprises the steps of (S10) acquiring eye gazing data indicative of a gazing direction of the driver; (S20) calculating coordinates of a preview point of the driver in a driving scene surrounding the vehicle based on the eye gazing data; and (S30) determining an intended trajectory of the driver based on the coordinates of the preview point.