Cross-Baseline Stereo Vision for Reliable Robot Safety Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing safety monitoring systems using stereo cameras for robots suffer from poor detection performance when objects are parallel to the camera baseline, leading to unreliable depth information and limited field of view, especially in open spaces.
Innovation Solution
A robot safety monitoring system with a stereo camera set comprising two cameras arranged in a T-shape or cross configuration, where each camera has different fields of view, allowing for stable detection by preventing one-to-one matching failures and enhancing object detection performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a stereo camera is used for safety monitoring, then the system is inexpensive and has a wide field of view, but detection performance degrades when objects are parallel to the camera baseline
Solution Approach 1:
The patent transitions from a single stereo camera system to a multi-stereo camera system with cameras arranged in multiple directions (front, rear, left, right). This dimensional expansion allows the system to capture objects from multiple baselines simultaneously, eliminating the detection degradation that occurs when objects are parallel to a single baseline. The multi-directional arrangement ensures that at least one camera pair will have a favorable baseline orientation relative to any given object.
2Ease of manufacture
If a line sensor is installed at an entrance to detect persons, then the method is cost effective, but it cannot be used in open spaces
Solution Approach 1:
The patent creates a universal safety monitoring system that can function in both open spaces and areas with defined entrances. The multi-stereo camera system with multiple baselines and wide field of view replaces the location-specific line sensor approach, providing cost-effective monitoring that adapts to various spatial configurations including open robot workspaces, gated areas, and transitional zones.
3Reliability
If a LIDAR sensor is used for monitoring, then detection is performed in a rotating manner, but detection is limited to a two-dimensional plane rather than three-dimensional space
Solution Approach 1:
The patent employs multiple stereo camera pairs arranged in three-dimensional space (front, rear, left, right orientations). Each stereo pair provides depth information along its baseline, and the combination of multiple pairs creates a three-dimensional monitoring volume. This multi-dimensional stereo vision system overcomes the 2D plane limitation of rotating LIDAR sensors while maintaining cost effectiveness.
4Adaptability or versatility
If a TOF camera is used for 3D monitoring, then data about three-dimensional space can be obtained, but the camera has a narrow field of view and is sensitive to surrounding brightness
Solution Approach 1:
The patent uses multiple standard stereo camera pairs to replicate the 3D detection capability that would otherwise require a single TOF camera. By distributing the 3D monitoring function across multiple conventional cameras with different baselines and orientations, the system achieves comprehensive three-dimensional coverage with wide field of view while avoiding the brightness sensitivity and narrow FOV limitations of TOF technology.
Data Source
AI summary
A robot safety monitoring system includes a stereo camera set configured to photograph a robot and a monitoring area around the robot and a processor connected to the stereo camera set. The stereo camera set includes a first camera including a plurality of first camera modules spaced apart from each other, and a second camera including a plurality of second camera modules spaced apart from each other. Based on a first direction, the first camera and the second camera have different fields of view. The processor is configured to set the monitoring area, determine whether an object is present in the monitoring area, and control the robot based on determining that an object is present in the monitoring area.