Robot Safety Vision Using 3D Skeleton Overlap Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional safety monitoring systems for robots require area sensors near the robot, restricting worker and robot movements, necessitating a solution that allows deceleration or stoppage without such sensors.
Innovation Solution
A safety vision device using human and robot three-dimensional skeleton estimation models to calculate overlapping areas based on joint point data and joint axis angles, enabling deceleration or stoppage of robots without area sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If an area sensor is installed near the robot to detect worker entry into the motion area, then safety monitoring capability is improved, but the mobility and operational flexibility of both worker and robot are restricted
Solution Approach 1:
The patent replaces the mechanical area sensor system with a vision-based detection system using cameras and skeleton estimation models. This substitution eliminates the need for physical sensors near the robot, thereby maintaining safety monitoring capability while removing restrictions on worker and robot mobility.
Solution Approach 2:
The patent introduces an external camera as an intermediary device positioned away from the robot to capture images for safety monitoring. This intermediary approach allows safety detection without requiring sensors in the robot's immediate vicinity, preserving operational flexibility while ensuring worker safety.
2Reliability
If an area sensor is used to monitor worker entry into robot motion area, then collision prevention is improved, but the system complexity and installation requirements increase
Solution Approach 1:
The patent extracts the safety monitoring function from the robot system itself and places it in an external vision processing system. By taking out the area detection capability from the robot's immediate environment and implementing it through external camera imaging and skeleton estimation, the patent reduces system complexity and installation requirements while maintaining collision prevention effectiveness.
3Reliability
If conventional area sensors are installed to detect worker position, then safety monitoring is achieved, but worker and robot movements are restricted
Solution Approach 1:
The patent transitions from two-dimensional area sensor detection to three-dimensional skeleton estimation based on camera imaging. By utilizing depth information and spatial coordinates from the vision system, the patent achieves accurate worker position detection without imposing physical constraints on movement, thereby maintaining safety monitoring while preserving movement freedom.
Data Source
AI summary
A robot decelerates or stops, without using an area sensor, when a worker enters a robot operating area. The safety vision device includes a human three-dimensional skeleton estimation model, a robot three-dimensional skeleton estimation model, an input unit for inputting a two-dimensional image of a worker and a robot captured by an external camera, and a distance and a tilt between the camera and the robot, an estimation unit that inputs, to the models, the two-dimensional image and the distance and tilt, to estimate three-dimensional joint point data indicating the three-dimensional coordinate values of the position of a joint point of the worker, and the angles of a plurality of joint axes included in the robot, and an approach determination unit that calculates a worker area and a robot area, and outputs an instruction to decelerate or stop the robot, depending on the degree of overlap of the areas.