Wearable Glasses for ADAS Using Deep Learning Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Advanced Driver-Assistance Systems (ADAS) are costly due to the use of multiple sensors and cameras, and require precise installation to provide accurate driving information, which can be compromised by errors in position and sensing direction, leading to inadequate driver assistance.
Innovation Solution
The use of assistance glasses equipped with a camera, acceleration sensors, and gyroscope sensors that employ a deep learning-based safe-driving information analyzing device to process visual-dependent driving images and sensor data, integrating convolution, detection, segmentation, and recognition networks to detect objects, lanes, and driving environments, and provide real-time collision, lane departure, and driver status warnings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple sensors and cameras are used for recognizing driving environment and driver status, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent combines multiple sensing functions (image capture, acceleration sensing, gyroscopic sensing) into a single wearable device unit that the driver wears, integrating what would traditionally be separate cameras and sensors into one consolidated system
Solution Approach 2:
The wearable device performs multiple functions simultaneously: capturing driving environment images, monitoring driver status through acceleration and gyroscopic data, and providing feedback through display and audio outputs, replacing the need for separate dedicated sensors for each function
2Measurement precision
If multiple sensors and cameras are installed at optimal positions, then measurement precision is improved, but device complexity and installation difficulty increase
Solution Approach 1:
The patent consolidates multiple sensing components into a single wearable unit that is installed once on the driver, eliminating the need for multiple separate installation locations and simplifying the installation process while maintaining comprehensive monitoring coverage
Solution Approach 2:
The wearable device is designed to be self-contained and portable, allowing the driver to wear it directly without requiring complex vehicle installation infrastructure, making the system easier to deploy and adjust
3Measurement precision
If sensors and cameras are installed on the vehicle, then surrounding monitoring is improved, but the system cannot accurately detect driver status
Solution Approach 1:
The wearable device is worn directly by the driver, allowing it to capture driver status information (through acceleration and gyroscopic sensors) and driving environment information (through the camera) simultaneously from the driver's own perspective, eliminating the information loss that occurs when vehicle-mounted sensors try to infer driver status
Solution Approach 2:
The wearable device acts as an intermediary between the driver and the vehicle system, directly capturing both driver physiological data and driving environment data, and transmitting this comprehensive information to the vehicle's processing system
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for providing safe-driving information via eyeglasses of a driver of a vehicle is provided. The method includes steps of: a safe-driving information analyzing device, (a) if a visual-dependent driving image, corresponding to a perspective of the driver, from a camera on the eyeglasses, acceleration information and gyroscope information from sensors are acquired, inputting the visual-dependent driving image into a convolution network to generate a feature map, and inputting the feature map into a detection network, a segmentation network, and a recognition network, to allow the detection network to detect an object, the segmentation network to detect lanes, and the recognition network to detect driving environment, inputting the acceleration information and the gyroscope information into a recurrent network to generate status information on the driver and (b) notifying the driver of information on a probability of a collision, lane departure information, and the driving environment, and giving a warning.