Automated inspection in physical industries is the process of performing automated and nondestructive testing (NDT) of products or automating routine inspections of infrastructure. Although this term has been around for decades now, it has seen significant activity recently, which is evident from the Lux Tech Signal below. In particular, we see that innovation activity in automated inspection systems (AIS) has taken off since 2010. In this blog, we will discuss different types of AIS and identify the key technological and market factors driving adoption.
Figure 1: The Lux Tech Signal for automated inspection systems
The working mechanism of AIS includes three main steps. The first step is the autonomous relative movement between sensors and test objects so that every individual test object or its parts can be inspected by the sensor. This can be done either by autonomously moving the sensor around stationary test objects, such as in drones inspecting a cell phone tower, or by moving the products around stationary sensors, such as in X-raying consumer packaged goods (CPG) on a conveyor belt. The second step is actual sensing, and the final step is identifying whether or not the measured value is a regular reading or an anomaly. In case of an irregular reading, which signifies a defect or aberration, the AIS flags it to human operators.
There are several methods to automate inspection in physical industries. Some of the most common ones are:
- Camera-based inspection: This is the most popular technique in AIS, which includes the use of cameras and computer vision algorithms to inspect products and infrastructure. Depending on the use case, the camera-based inspection may use different types of cameras, including RGB cameras, stereoscopic 3D cameras, hyperspectral cameras, thermal cameras, and infrared (IR) cameras. One of the main reasons for the popularity of camera-based inspection is that the cameras are small in size, low in power consumption, easy to set up, and inexpensive. In addition, the rapid development of computer vision algorithms to more accurately and robustly assess objects will further fuel the adoption of optical inspection across various industries.
- Acoustic inspection: This technique includes the use of acoustic microscopes that emit ultra-high-frequency sound waves for inspection. Unlike camera-based inspection, which faces difficulties with occlusion, highlights, and shadows, acoustic inspection can be used to identify surface as well as subsurface flaws and cavities in metal structures. However, as discussed in our previous insight, the use of acoustic inspection is limited to applications that generate high-decibel sound, preferably from cyclic motion, such as in wind turbines.
- Nonoptical electromagnetic radiation-based inspection: This technique includes the use of X-rays and radars to inspect objects. For example, a ground-penetrating radar (GPR) uses radar pulses in the microwave band to image the subsurface of earth and metals. On the other hand, X-ray-based inspections are more popular in the food industry for applications in foreign body detection, including bone fragments in processed meat that move on a conveyor belt.
- Electrical and magnetic field-based inspection: These types of AIS include inducing electric current, magnetic fields, or both inside test objects and using voltmeter or magnetometer probes mounted on a robotic arm to observe the field response at different parts of the test object. The most popular type of electromagnetic inspection is automated eddy current testing, which is used in surface and tubing inspection. Examples of surface inspection include detecting cracks on airplanes and leakage in oil and gas reservoirs. Similarly, examples of tubing inspection include detecting and sizing pits in components like heat exchangers.
Besides these four, there are several other types of AIS that use mechanical or chemical sensors, including acceleration, vibration, pressure, and gas sensors for inspecting products and infrastructure.
Driving factors of AIS
There are two groups of factors responsible for the growing activity in AIS – technology-driven factors and market-driven factors.
- Technology-driven factors: One of the biggest drivers of AIS is the use of AI tools like machine learning algorithms to assess anomalies. Traditional AIS use rules-based techniques, which are manually programmed into the system in the form of if-then-else statements, to differentiate anomalies or defects from regular events. Such rules-based inspection systems can only capture anomalies that have already been identified and hard-wired into the system. While the rules-based techniques perform with satisfactory results in controlled environments like automotive manufacturing, where the chances of new kinds of anomalies are minimal, they are less useful in more complex environments like quality inspection of food as it passes through different stages of the supply chain. Machine learning algorithms, on the other hand, define anomalies by learning from training data. Therefore, instead of a programmer hardwiring rules, the algorithms create the rules themselves, which enables them to model and identify anomalies in new and more complex environments. As an example, UVeye claims that with the use of deep learning algorithms, its AIS is able to identify anomalies in vehicles that it never detected before.
- Market-driven factors: Although most inspection tasks are still performed manually, the market demand for AIS is expanding for several reasons. First, the increasing wage demands of workers and their reluctance to perform monotonous inspection jobs have both raised issues for employers. In addition, some jobs, such as inspection of tall infrastructure (telecommunication towers, bridges, wind turbines, etc.), are highly risky. With the rise in worker monitoring and safety practices in industries, such jobs can require additional expenditures if done manually. Furthermore, there are systems like steam valves that are impossible for human workers to inspect manually. AIS companies are beginning to develop solutions to address many of these challenges. For example, Nearthlab deploys aerial drones to automatically inspect wind turbine blades, which is otherwise a risky and time-consuming job for human workers. Overall, AIS offer a scalable solution to improve efficiency and accuracy and reduce the need for manual intervention in many inspection tasks.
- Companies should note that both technology innovation and market demand will continue to make AIS more feasible in many use cases.
- Among different types of AIS, camera-based inspection will continue to gain the most adoption because of the convenience and accessibility of cameras and rapid advancements in underlying technologies.
- Due to the benefits, such as safety, scalability, and efficiency, AIS will replace or augment manual inspection in many industries.
- However, given the high upfront installation costs and ongoing maintenance, a clear return on investment (ROI) will remain the key driver of what tasks see the quickest adoption of AIS.
- Blog: What to Expect from AI in 2020
- Download the Executive Summary: Key Players and Business Models in the Connected Car