Wearable Obstacle Detection Using Normal Map Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing visual aid devices for individuals with low vision are not sufficiently accurate and versatile in detecting obstacles, often leading to false detections and limiting the autonomy of users.
Innovation Solution
A method and system using a distance sensor mounted on a wearable device to obtain a normal map, apply orientation correction, and segment the map into obstacle, floor, and lateral wall zones, utilizing various sensors like ToF, LIDAR, and RGB-D cameras, with IMU and visual odometry for accurate obstacle detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional distance sensors are used for obstacle detection, then the device can detect obstacles, but the detection accuracy is insufficient and false detections occur frequently
Solution Approach 1:
The patent segments the detection space into multiple zones (lateral wall zone, floor zone, obstacle zone) based on surface normal parameters. By dividing the detection task into spatial segments, the system can apply different analysis criteria to each zone, significantly improving detection accuracy and reducing false positives compared to uniform detection approaches.
Solution Approach 2:
The patent introduces orientation dimension by analyzing surface normal parameters in addition to distance. Instead of relying solely on distance measurements, the system incorporates the directional information of surface normals to distinguish between different types of surfaces (floor, wall, obstacle), thereby enhancing detection precision and reliability.
2Measurement precision
If the sensor coordinate system is not corrected for orientation, then the processing is simpler, but the detection results are inaccurate when the user changes head inclination
Solution Approach 1:
The patent performs preliminary orientation correction by transforming the sensor coordinate system to align with the world coordinate system before conducting obstacle detection. This preliminary action ensures that all subsequent detection operations are performed in a consistent reference frame, improving accuracy without requiring complex real-time adjustments during detection.
Solution Approach 2:
The patent introduces an intermediate coordinate transformation step that acts as a mediator between the sensor's local coordinate system and the world coordinate system. This intermediary transformation layer simplifies the overall system by providing a standardized reference frame, making the detection algorithm more robust to head inclination changes while maintaining manageable complexity.
3Adaptability or versatility
If simple obstacle detection methods are used, then the device complexity is lower, but the system cannot provide accurate spatial segmentation of the environment
Solution Approach 1:
The patent implements spatial segmentation by dividing the environment into distinct zones (lateral wall zone, floor zone, obstacle zone) based on surface normal parameter analysis. This segmentation approach enhances adaptability by enabling the system to recognize and respond to different spatial configurations, while the algorithmic complexity is managed through efficient geometric computations.
Solution Approach 2:
The patent applies local quality analysis by examining surface normal parameters at different spatial locations to determine zone classification. Each region of the detected environment is analyzed with specific criteria appropriate to its location and orientation, enabling accurate spatial segmentation without requiring uniformly complex processing across the entire field of view.
Data Source
Figure 1A~1B
Figure 2
Figure 3
AI summary
A method and system for detecting obstacle elements is presented in the present disclosure, the detection performed with a visual aid device, the device comprising a distance sensor, the sensor being mounted in a wearable device to be worn by a user. Several embodiments of the detection method and system allow the further detection of different types of obstacles, in order to map and signal them to a user with low-vision.