Industrial inspections are typically labor-intensive and conducted on a time-based schedule. The collection of data is generally manual, time-consuming and involves humans being exposed to high-risk environments. For example, ropes teams hang from high places, inspectors enter confined spaces and workers often are exposed to severe elements. These inspections result in suboptimal data collection based on a human’s ability to work in such environments.

Data collection is subjective, inconsistent and prone to human error. Inspections are conducted on a periodic basis, and sometimes the find rate of time-based inspections is less than 2%, making more than 98% of the inspections unnecessary. In other instances, the inspection interval is too long, and adverse incidents occur as a result. For example, to inspect some assets in the oil and gas industry, such as flare stacks, operators must take them out of service, causing lost revenue.

Once inspection data are collected, value must be extracted from it. This process also is manual and not integrated, leaving findings disorganized and subject to interpretation. These findings can vary from inspector to inspector or even from one day to the next. In the process of transferring and communicating inspection results for paper or PDF reports, data can be lost. Additionally, firms often delete much of the data after the report is finished, which limits the ability to learn from historical records. Overall, reporting can take six months or more after the inspection is completed.

Platform detects anomalies, generates alerts

To address these challenges, some companies are using unmanned aerial vehicles (UAVs) or AUVs. During a subsea inspection, for example, many companies require a support vessel and crew to command ROVs. This process is usually expensive and weather dependent. During aboveground inspections, UAVs can collect thousands of images, but inspectors need to manually examine these data to develop a report, which doesn’t save much time. The true benefits of robotic inspection derive from the ability to efficiently maximize the added data that are collected over time with a smaller logistical footprint.

Avitas Systems, a GE Venture, is taking a more systematic approach to inspections by collecting data autonomously and fusing those data into an advanced analytics platform, where artificial intelligence (AI), physics models and algorithms combine to automatically detect asset anomalies and generate alerts.

To develop these solutions, Avitas Systems brings together a team of subject matter experts across diverse technologies and businesses, including engineering, analytics, computer vision, flight operations and GE’s Global Research Center for the customization of these technologies.

Avitas Systems uses a variety of robotics, including UAVs, AUVs, surface robots such as crawlers and customized sensor technology, to make its services crossindustry, targeted and thorough. By incorporating robotics in data collection, the company can prevent inspectors from performing high-risk tasks and provide more consistent inspections to better detect asset defects earlier and faster, which means communities surrounding industrial assets are safer, too. All the while, minimal machine downtime is required, so companies don’t have costly turnaround time.

Advanced 3-D modeling, AI

What makes robotic data collection with Avitas Systems more efficient is the company’s advanced 3-D modeling and unique integration of AI to enable its analytics. The company’s data collection process improves the quality and consistency of inspection insights, which customers often can see livestreamed. Inspectors can select exact points of inspection (POI) on digital 3-D models of entire assets.

Users simply select POI on the model, change the perspective of the model to define the sensor angle and indicate the size of the resolvable defect by extending the POI, which translates into a robot’s standoff distance. This point-and-click method reduces the inspection planning time from hours to minutes, with a more targeted approach. Avitas Systems autonomously converts this 3-D modeling, integrated with existing customer requirements such as no-fly zones, into safe and precise paths that robots follow for data collection.

These paths can be repeated, which allows increased efficiency and the ability to detect changes in an asset over time. Change detection produces historical data in the form of images that sensors capture, including RGB, infrared and ultraviolet imaging, stored on the company’s digital, cloud-based platform. The platform centralizes the data, unlike many other companies, and allows archival searches of inspection records as the data grow.

Deep learning models are stored on the original Avitas Systems AI Workbench. (Source: Avitas Systems)


Predictive analytics

In addition to data warehousing, the platform notably includes predictive analytics so inspections can be planned according to risk, as opposed to time intervals. Risky assets receive increased attention, thus improving safety. Avitas Systems fuses manual and autonomous inspection data, existing asset performance data, external data sources (e.g., weather) and new inputs from subsequent inspections. Advanced algorithms then detect asset defects and anomalies automatically. Defects and anomalies vary across the industry, including flare stack damage or subsea gas bubbles.

Avitas Systems can quickly and more accurately determine the likelihood a defect will lead to failure in a few hours instead of several weeks and at a reduced cost. Automated defect recognition means inspectors no longer need to manually peruse disparate datasets across different teams. As more data are ingested from diverse sources, the deep learning models stored on Avitas Systems’ original AI Workbench retrain for smarter actionable insights.

The platform uses predictive analytics to recommend targeted inspection scheduling and planning, which significantly improves accuracy and enables earlier resolution of potential issues. The recommendations and networkwide risk maps for assets are displayed in the company’s customer-focused, web-based interface with accessible dashboards based on user type. The advanced reporting tools available in the interface integrate with existing management and reporting systems.


With the company’s hybrid of technologies, human perspective complements the dexterity of advanced robotic technology and analytics. Avitas Systems is making inspectors’ jobs easier and more efficient. This first inspectionas- a-service solution and the fundamental algorithms it involves can apply to multiple industries, including oil and gas, electric power and transportation. Avitas Systems is partnering with market leaders in robotics and AI to expand its platform globally. By partnering with Kraken Robotics, for example, Avitas Systems will be able to incorporate sensor technology for subsea inspection across the oil and gas, offshore renewable energy and shipping industries.