Unleashing the full potential of vision technology in automated systems requires effective design, testing and collaboration.
Today’s industrial machine vision systems use high-resolution cameras, image processing algorithms, and artificial intelligence to enhance quality control, automate inspections and optimize production processes with precision and efficiency.
The implementation of vision systems spans applications like 2D barcode scanning, pattern and character recognition, measurement and error proofing. Over the years, the introduction of these systems for use in an industrial setting has been the subject of much debate.
Although there is potential for vision to enhance many aspects of manufacturing and assembly operations, the practical execution of these systems often falls short of the expected goals. The misapplication of vision tends to result from a failure to align business goals with technical goals, leading to architecture that does not meet the needs of the organization.
Understand the Purpose and Expected Outcomes
It’s important not to lose sight of why a vision system was selected in the first place. A business goal of zero defects to the customer is one path that leads to vision as the technical solution. However, recognizing that there could be other solutions is important if the initial vision solution does not deliver. In this example, the goal of zero defects would require a vision success rate of 100 percent, but a first-time success rate of 100 percent is extremely difficult to achieve in an industrial environment.
This disconnect between business goals and the resulting system architecture can become a major source of hardship for the implementation team. I have spoken to plant managers who have expressed regret over allowing cameras into their facilities as the actual camera failure rate caused an unacceptable amount of download time and lost production. It is important to note that dissatisfaction was not a result of the quality or purpose of vision systems, but from a lack of understanding of the limitations of the technology combined with unrealistic expectations of performance.
The Importance of Effective Implementation
The treatment of these systems as a general project requirement as opposed to meeting specific business or technical goals creates difficulties for OEMs. Tight schedules can create a hurried development environment, resulting in poor specifications and incomplete testing. Since many OEMs lack specialized vision expertise and rely on supplier support, implementation without proper reliability or repeatability testing is a serious risk. Consequently, substandard systems require significant rework before they can effectively support production once installed at the customer site. The system will then fall short of the actual business requirements, lacking in functionality, robustness or maintainability. Also, operator acceptance procedures and programming are frequently left for the customer to develop, which compromises overall system effectiveness and robustness. The responsibility of testing and final implementation can end up falling on the controls and process engineer, leading to re-engineered applications and unanticipated problems that require extensive troubleshooting.
The Role of Design, Testing and Optimization
The success of vision systems relies on proper up-front design, testing, application optimization and detailed requirements documentation. Despite engineers’ efforts to find cost-effective solutions that meet requirements, poor execution at the OEM level can undermine those efforts. High-quality cameras alone cannot compensate for suboptimal design and testing processes. Extensive fine-tuning is necessary after installation, but due to time constraints this stage is often rushed, leading to frustrations and setbacks. This rushed approach contrasts with the need for gaining a comprehensive understanding of the functionality and performance of the vision systems and their limitations. User Acceptance testing should include detailed simulation, testing and analysis of results. Risk assessment should precede any modifications to camera parameters after buyoff, and these modifications should only be implemented after thorough vetting.
The Dilemma: Can a reasonably priced vision system be implemented with a high success rate, good maintainability, easy adjustability and programmability? The answer is: that is depends on how the alignment between business goals and technical ability is understood. Misapplication of actual requirements that fall outside the capability of the technology can be catastrophic. A fault tolerant, error free system can be achieved, but comes with a cost in terms of time and financial investment that some manufacturers may hesitate to commit.
Camera Limitations: Another vital aspect to consider is the understanding of failure rates. Adjustments made to a vision system with the aim of enhancing performance may take weeks to prove effective, even at failure rates as high as 1 in 10,000 pieces. Any modification could unintentionally increase the number of rejects, and it is only after running thousands of parts that the impact of any modification can be adequately evaluated.
A common application is the use of cameras to read barcodes commonly used for tracking. Fixed cameras can never attain a 100 percent read rate, resulting in undesired production line stoppages. Obstruction (dust, dirt, coolant etc.), poor lighting, camera misalignment, or damaged barcodes can also cause inaccurate readings. To compensate for these challenges, backup handheld scanners are often deployed at various stations. While smartphones are readily used to point to the camera as the problem, the real issue often lies in the image quality itself. Simulation is a valuable tool that can be used behind the scenes to validate solutions before implementation or after a change, but simulation cannot entirely replace live testing. Therefore, any tweak to a camera should prompt the implementation of a bypass until the new parameters are thoroughly vetted.
Balancing Cost and Performance
Creating a vision system with a high success rate, straightforward maintenance and flexibility is feasible. However, company leadership must understand that the benefits of well-designed, easily adjustable and programmable systems are critical to ensure both business and technical needs are met. By striking a balance between cost, performance and the system’s ability to adapt and remain maintainable, manufacturers can still minimize downtime, optimize production processes and enhance overall operational efficiency.
Implementing vision systems in industrial environments presents a number of challenges that must be overcome for successful integration. Effective implementation requires careful consideration of factors such as design, testing, optimization, and alignment of expectations. The role of OEMs in delivering high-quality, fully functional systems cannot be understated, and close collaboration between OEMs and plant controls engineers is essential. While limitations exist, especially in barcode cameras, a proper understanding of the system’s purpose and limitations can aid in identifying alternative solutions or mitigating risks. By acknowledging the importance of thorough testing, comprehensive analysis of failure rates, and the need for adaptability, manufacturers can pave the way for the successful implementation of vision systems. The journey towards achieving high-performing vision systems is probably longer than you plan for at the start of a project, but the long-term benefits in terms of productivity, quality, and operational efficiency make it a worthwhile endeavour.
