Advancing Factory Automation with Smart 3D Inspection
IIoT and The Move Toward Smart: Leading a Global Trend
The Industrial Internet of Things (IIoT) is a term used to describe the importance of interconnectivity in relation to factory automation. IIoT summarizes the complex interrelationship of smart hardware feeding actionable data over a network to drive deep learning systems toward optimizing factory output. This paradigm shift is causing fundamental advancements in the way products are manufactured and delivered.
Inline quality control is an essential part of the IIoT equation. Devices such as 3D smart sensors are driving factory automation forward with their ability to measure critical part tolerances, communicate seamlessly with factory robots and networks, and deliver real-time data in order to increase factory efficiency and achieve 100% quality control and material optimization in high-speed, high-volume inline production environments.
The Power of 1%
To put the magnitude of this “smart revolution” in perspective, a mere 1% improvement in factory efficiency and utility operating costs in these five industries––Oil & Gas, Power, Healthcare, Aviation, and Rail––would add up to $276 billion in savings over 15 years. North American industries alone stand to achieve $500 million in annual savings.
3D Smart Sensors, IIoT and FactorySmart®
3D smart sensors advance smart factory automation by offering a tight integration of non-contact 3D scanning and measurement to make critical decisions and communicate this to factory networks –– all within a single package. LMI designs and manufactures Gocator, a 3D smart sensor for the quality inspection market offering this all-in-one capability powered by web browser-based technology for configuration and monitoring by mobile devices.
Understanding the Markets for 3D
The world of 3D technology is made up of two primary markets: (1) Metrology, and (2) Inspection.
Metrology is carried out in an offline measuring environment that is clean and controlled –– typically a dedicated measuring room. Solutions used for metrology applications are largely based on the use of CMM (coordinate measurement machines) and contact-based touch probes.
These solutions are used to perform Geometric Dimensioning and Tolerancing (GD&T) analysis and reporting for first article type parts that can take hours of setup and measurement.
Of greater interest and opportunity is the inspection market. This market requires an entirely different set of 3D capabilities. Inspection applications are typically inline or at-line, and carried out in harsh factory environments that introduce variations such as ambient lighting, vibration, temperature fluctuation, dust, water, oil, etc. Furthermore, data acquisition and measurement (i.e., cycle and tact times) are very short for quality inspection applications. Laser-based solutions are the natural choice in this market, where the sensor must deliver accurate pass/fail decisions on manufactured parts being scanned continuously at factory speeds. Traditional metrology has the luxury of time whereas inline inspection does not. Metrology machinery can be calibrated and recalibrated whereas inline sensors must be installed to deliver pre-calibrated, reliable output for years with no chance for recalibration later.
Fig 1. Inline inspection requires a combination of high-speed, resolution and repeatability.
Key Challenges of Inline Measurement
For inline measurement applications, the top most drivers that sensors must address are resolution, repeatability and reliability. First, sensors have to scan with sufficient resolution to identify and confirm critical features are meeting manufacturing tolerances. Resolution must be at least twice as high as the feature to ensure detection. Second, sensors must deliver the same measurement for the same part with minimal variation. This high repeatability requires the underlying sensor to have low “noise” and not be adversely affected by external factors like temperature or ambient lighting. For measurement repeatability, resolution must be at least 4-8x as high as the feature size to ensure enough data points for repeatable results. Third, the sensor must be reliable –– delivering the same performance on day 1 as on day 1,000. This means the internal electro-optical-mechanical design must be robust, maintaining its factory calibration performance over time with no degradation in the optical path.
A 3D smart sensor like Gocator is specially designed to achieve these critical metrology goals in addition to keeping up with factory production speeds.
Shape-Based 3D Inspection
One of the key advantages of 3D over 2D is its ability to capture shape. In a 3D smart sensor, tools are available for measuring geometry such as the diameter of a hole located on the side of an object. Shape-based inspection is not affected by the color of the object whereas 2D relies solely on the contrast or “color” in an image. With all-in-one processing, a 3D smart sensor allows users to acquire 3D point clouds along with 2D intensity data, then find edges based on shape or contrast and produce detailed feature and volumetric measurements that are collected for pass/fail control decisions.
Fig 2. Shape-based measurement is a key capability that is unique to 3D technology.
Tag and Track Control
In addition to 3D inspection based on shape, 3D smart sensors offer built-in processing to carry out tagging and tracking parts. This allows pass/fail decisions to be stamped by the time and position of the part on a conveyor for downstream coordination with ejection or sorting hardware. When the part arrives at a sort or reject bin, the pass/fail decision determines if dedicated eject hardware is activated or not. This requires the sensor to manage a queue of hundreds of parts in its memory that are individually tracked and processed after inspection is completed.
Standard 3D sensors deliver fixed functionality in scanning. Smart sensors extend scanning by adding built-in measurement and communication support. But in the end, what you purchase is what you get. A new breed of smart sensor, however, opens the possibility for developers to add custom measurement algorithms that run in the sensor by replacing the embedded firmware.
With Gocator Development Kit (GDK), you can embed your own custom measurement tools into the sensor firmware––with the same functionality as native built-in tools––while benefiting from the ease of use of a web-based user setup that leverages built-in 3D visualization and drag and drop workflow.
Fig 3. The Gocator Development Kit (GDK) allows developers to create their own unique inspection solution by adding custom measurement algorithms to Gocator.
Data Processing Acceleration
For applications like electronics inspection where a standard one-second cycle time is required, data processing acceleration is necessary for end users to meet stringent QC and throughput specifications.
For this reason, LMI offers GoX, a Windows PC application that allows the user to add the data processing power of a PC to their inspection solution. The application significantly increases processing speed and reduces cycle times, removes memory limitations, and allows the user to handle large 3D point clouds for measurement and inspection in the required cycle time. When a sensor cannot meet timing for an application, the GoX is used to seamlessly accelerate the sensor’s processing to offer 4-10X speed increases.
Virtual Sensor Capability
Reviewing recorded data from factory production lines to optimize measurement parameters is a common requirement to optimize inline inspection performance . For this purpose, Gocator Emulator provides users with a "virtual" sensor that can be used as a safe, offline testing environment to ensure algorithms are reliable and performing well for inline production environments.
Emulator gives programmers the ability to determine issues with current sensor configurations, test custom algorithms, and design and test improvements in an offline environment prior to deploying their solution onto an actual sensor.
The vast majority of inline inspection applications require multiple sensors, set up in a variety of orientations to provide 3D point clouds that cover the critical aspect of an object and eliminate any occlusions that might be present.
For this reason, Gocator can be configured in a range of multisensor layouts to fit individual inspection needs. Both dense and sparse multisensor configurations are supported. Dense configurations result in merging all sensor data streams into a single 3D point cloud whereas sparse networks result in processing sensor data individually (eg., gaps between sensors are not filled in). In either case, calibration methods are built-in to support transforming data from each sensor in a multisensor network to a common world coordinate system.
LMI’s Master Hub allows for the seamless networking of the smart 3D sensors. The Hub is a dedicated solution for distributing power, data synchronization and laser safety in a multi-Gocator network. Masters allow users to easily scale from a single sensor up to a 24-sensor system, provide synchronization within 1 μs accuracy, and feature all-in-one cabling to carry data directly between sensors and a network switch. Synchronization data is broadcast to all sensors includes a timestamp, encoder stamp, and the status of direct inputs wired to the Hub
Fig 4. Master Hubs allow users to easily create fully synchronized Gocator sensor networks.
Gocator 3D smart sensors provide all the necessary features and functionality to drive smart 3D inspection. From 3D+2D inspection, to sensor customization, acceleration, emulation and multi-sensor networking, the 3D smart sensor is designed for the demands of IIoT and the modern factory.