US20250095388
2025-03-20
Physics
G06V20/597
The patent application describes a method and system for updating user predictive mental response profiles, particularly for drivers of autonomous vehicles. The system uses imaging sensors to capture images of a user's eyes and analyzes these to identify eye dynamics signal patterns that precede abnormal events. By associating these patterns with specific events, the system updates the user's response profile, allowing processing units to predict and respond to imminent abnormal events. This technology aims to enhance the interaction between human users and partially autonomous systems.
The development of autonomous vehicles has advanced rapidly, integrating multiple disciplines to address various challenges, particularly in human-machine interaction. Autonomous systems, even when fully automated, may require human oversight. The transition period towards fully autonomous vehicles presents challenges such as managing driver attention and preventing passive fatigue due to low cognitive load. Conversely, high cognitive workload can also degrade performance. Designing systems that ensure drivers understand the capabilities and limitations of these vehicles is crucial for effective human-machine collaboration.
The invention offers a method for updating driver response profiles by capturing and analyzing images of their eyes. The process identifies eye dynamics signal patterns that occur before abnormal driving events. These patterns are used to update the driver's response profile, enabling vehicle control systems to anticipate and react to potential issues based on real-time eye dynamics data. This proactive approach aims to improve safety and efficiency in autonomous vehicle operations.
A system is proposed comprising processors that receive images from eye-monitoring sensors, analyze them for eye dynamics patterns, and update driver response profiles accordingly. This system can predict imminent abnormal events by leveraging updated response profiles, initiating appropriate actions within the vehicle's control systems. The invention also includes a non-transitory computer-readable medium storing software instructions for executing these methods.
The patent outlines optional features where additional eye dynamics patterns are identified after abnormal events to validate initial predictions. This involves analyzing images post-event to confirm the accuracy of pre-event signal patterns. If certain patterns occur within a specific time window after an event, updating the driver's response profile may be forgone, ensuring only relevant data influences future predictions.