Driver Turn Intention Detection Using Gaze and Odometry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining a driver's intention to turn are unreliable, especially when the route is unknown or navigation information is unavailable, leading to potential incorrect turn indications and compromised road safety.
Innovation Solution
A device and method utilizing an interior camera with a wide field of view, combined with environmental data from units like lidar and radar, and odometry data, to create heat maps and determine the driver's focus and turning intentions, ensuring accurate activation of turn indicators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If navigation information is used to determine turning intention, then automation is improved, but reliability deteriorates when route information is unavailable
Solution Approach 1:
The patent introduces an intermediary system that combines navigation data with direct driver behavior observation. When navigation information is available, it serves as a preliminary indicator, but the system mediates this with camera-based gaze and head position analysis to confirm actual turning intention, ensuring reliability even when navigation data is incomplete or unavailable.
Solution Approach 2:
The system implements feedback loops where the processing unit continuously monitors driver behavior (gaze direction, head position) and compares it with navigation information. This feedback mechanism allows the system to verify or correct turn indications, maintaining high reliability by using real-time driver behavior as a validation layer over automated navigation data.
2Reliability
If multiple data sources (camera, environment capture unit, odometry) are combined, then reliability is improved, but device complexity increases
Solution Approach 1:
The processing unit serves multiple functions: it processes camera images for gaze detection, integrates environment capture unit data for contextual awareness, analyzes odometry for vehicle state, and synthesizes all this information for turn intention determination. This multi-functionality consolidates what could be separate complex systems into a single coordinated processing unit.
Solution Approach 2:
The patent merges multiple data streams (visual data from camera, environmental data from capture units, motion data from odometry) into a unified analysis framework. The processing unit combines these diverse inputs to form a comprehensive assessment of driver intention, reducing the need for separate independent systems and managing complexity through integration.
3Ease of operation
If interior camera with wide field of view is used, then ease of operation is improved, but measurement precision of driver focus deteriorates
Solution Approach 1:
The processing unit segments the wide field of view into multiple regions of interest, particularly focusing on areas where the driver's eyes and head are most likely to be positioned. By dividing the broad capture area into zones with different analysis priorities, the system maintains comprehensive coverage while applying enhanced precision measurements to critical areas.
Solution Approach 2:
The system applies different processing qualities to different parts of the image data. In regions where the driver's face and eyes are detected, higher measurement precision is applied for gaze and head position analysis. In other areas of the wide field of view, standard processing suffices, optimizing the balance between comprehensive detection and precise measurement.
Data Source
AI summary
A device and a method determine an intention of a driver to turn. A sequence of images of an area of the vehicle interior is captured and processed to determine a focus area of a driver outside the vehicle. The odometry data is processed to determine the possibility of the vehicle turning towards the determined focus area. The environmental data is processed to determine whether odometry data that enables turning has another cause. The intention to turn is determined based on the results of processing the image data, the odometry data and the environmental data.