GB/T 45501-2025Industrial robots - Three-dimensional vision-guided system - General technical requirements (English PDF)

Foreword

GB/T 45501-2025 was issued on 28 March 2025 by the State Administration for Market Regulation and the Standardization Administration of China, and has been in force since 1 October 2025. The /T in the designation marks it as a recommended national standard.

This document was drafted in accordance with the rules given in GB/T 1.1-2020 Directives for standardization - Part 1: Rules for the structure and drafting of standardizing documents.

Attention is drawn to the possibility that some of the elements of this document may be the subject of patent rights. The issuing body of this document shall not be held responsible for identifying any or all such patent rights.

This document was proposed by the China Machinery Industry Federation and is under the jurisdiction of the National Technical Committee on Robot Standardization (SAC/TC 591).

The drafting organizations include Mech-Mind (Beijing) Robotics Technology Co., Ltd., Beijing Research Institute of Automation for Machinery Industry Co., Ltd., Hangzhou Hikrobot Co., Ltd., Tsinghua University, AUBO (Beijing) Intelligent Technology Co., Ltd., SIASUN Robot and Automation Co., Ltd. and other companies and institutes.

1 Scope

China's national standard for the 3D vision systems that tell an industrial robot where a part actually is, so that it can pick, place or assemble objects that do not arrive in a fixed position. A robot on its own repeats a taught path; with a 3D camera and the software behind it, it can take parts from a bin, depalletize mixed cartons or fit a component whose pose varies from one cycle to the next. The weak points of such systems are well known to integrators: the calibration between camera and robot, the drift that builds up as the installation runs and ages, and performance figures quoted by suppliers under conditions nobody can reproduce. The standard answers these by defining a reference architecture with a device layer and a control layer made up of seven modules, by setting basic and functional requirements covering configuration, hand-eye calibration, simulation, the operating interface, logging, drift self-correction, integration over OPC UA and maintenance, and by giving recommended performance indicators together with the test method for each one. It is written for the companies, universities and research institutes that develop and apply 3D vision guidance for industrial robots, and gives buyers and suppliers a common basis for specifying and checking a system.

This document specifies the architecture, basic requirements, functional requirements, performance requirements and other requirements of three-dimensional vision-guided systems used in the field of industrial robots, and describes the test methods for the performance indicators.

This document is applicable to guiding enterprises, universities, research institutes and other relevant organizations in the research, development and application of three-dimensional vision-guided systems for industrial robots.

2 Normative references

The contents of the following documents constitute indispensable provisions of this document through normative references in the text. For dated references, only the edition corresponding to that date applies to this document. For undated references, the latest edition (including all amendments) applies to this document.

GB/T 12642-2013 Industrial robots - Performance criteria and related test methods.

GB/T 12643 Robots and robotic devices - Vocabulary.

GB/T 33863.8 OPC unified architecture - Part 8: Data access.

GB/T 39005 General technical requirements for vision integration systems of industrial robots.

3 Terms and definitions

For the purposes of this document, the terms and definitions given in GB/T 12643 and GB/T 39005 and the following apply.

3.1 three-dimensional vision sensor - a device that combines optics, algorithms and electronics to calculate depth information of a scene or object using various principles, such as stereo vision, structured light or time of flight. Note: it is commonly used in industrial automation, robot navigation, quality inspection, medical imaging, virtual reality and augmented reality.

3.2 three-dimensional vision-guided system - a technical system that uses three-dimensional vision perception to obtain spatial information about a target object or environment (such as shape, position, orientation and depth), processes and analyses that information by algorithms, and finally generates control commands to guide a mechanical device (such as a robot or automated equipment) to carry out precise operations.

3.3 hand-eye calibration - the process of determining the exact position and orientation of the vision system relative to the robot by mathematical modelling and optimization, so as to achieve accurate perception and operation. Note: it is commonly used in robot vision systems to ensure that image data are consistent with actual spatial motion.

3.4 system drift self-calibration - a technique or method that, when the performance of the system drifts because of long-term operation, changes in the external environment or ageing of internal components, dynamically adjusts and corrects the system by means of internal algorithms or sensor data, so as to restore its accuracy and stability.

3.5 measurement accuracy - the closeness of a measurement result to the true value.

3.6 speed of vision processing - the time required by the vision module or system from the start of image acquisition to the output of the result. Note: the intermediate process includes image acquisition, data transmission and algorithm processing.

3.7 repeatability of recognition - the consistency of the results when the same system or method recognizes the same target several times under the same conditions.

3.8 recognition success rate - the proportion of cases in which a system or method can accurately recognize the target object or feature under specific conditions. Note: it is usually expressed as a percentage and reflects the correctness and reliability of the recognition process.

3.9 guidance positioning accuracy - the degree of precision with which the guidance system guides the target to the intended position. Note: it reflects the positioning capability of the system.

4 Abbreviations

The following abbreviations apply to this document.

AP: accuracy of pose. API: application programming interface. DNN: deep neural networks. EtherNet/IP: Ethernet/Industrial Protocol. FR: frame rate. MTBF: mean time between failures. MTTR: mean time to restoration. OPC UA: Open Platform Communications Unified Architecture. RP: repeatability of pose. SVM: support vector machine. TCP/IP: Transmission Control Protocol/Internet Protocol.

6 Basic requirements

The three-dimensional vision-guided system shall meet the following basic requirements.

a) The processing results, speed and so on meet the requirements of the actual production system.

b) The system adapts to changes, including changes in working distance, in the recognized object and in external systems, where: 1) changes in working distance include changes in the spatial position of the recognized object and of the three-dimensional vision sensor; 2) changes in the recognized object include changes in its state of motion, colour, surface material and shape; 3) changes in external systems include changes in temperature, humidity, cleanliness class, ambient light intensity and vibration intensity.

c) The system can interact with mainstream industrial robots and business management systems, and with operators through the human-machine interface.

8 Performance requirements

Clause 8 gives recommended values for seven performance indicators of the system - acquisition frame rate, number of acquired points, measurement accuracy (expressed relative to the working accuracy of the robot), speed of vision processing, repeatability of recognition in position and orientation, recognition success rate, and guidance positioning accuracy of the robot tool centre point in position and orientation - the last two stated for vision-guided assembly scenarios. A note records that guidance positioning accuracy also depends on the performance of the robot itself and on the result of hand-eye calibration.

10 Test methods

Clause 10 gives a test method for each performance indicator of Clause 8. Each method states the test conditions, the test steps and the formula by which the result is calculated, with repeated measurements averaged; the measurement accuracy test uses a ball bar with diffusely reflecting spheres, evaluates the sphere spacing error by least-squares sphere fitting and repeats the measurement at several positions covering the working space of the sensor.