Azure Applied AI Services Accelerates Development of AI Solutions to Help Businesses Take Flight
At any given time, turnaround coordinators at German airline Lufthansa CityLine have their eyes riveted on monitors displaying more than half a dozen video feeds of planes parked at gates around the airport. The job of coordinators is to ensure that planes are unloaded, refueled, cleaned, restocked and reloaded so that every passenger reaches their destination safely, on time and with their luggage.
The minutes wasted here and there in the turnaround process can add up and cost airlines millions of dollars a year. As many in the industry note, planes only make money in the air.
“Think of a pit stop in an auto race, and it’s pretty much the same thing that happens when flipping a plane,” said Philipp Grindemann, head of business development and project management for Lufthansa CityLine. “All processes must be punctual, fast and lean.”
Lufthansa CityLine is a subsidiary of Lufthansa, one of the world’s leading airline groups with a network that covers the whole world. Lufthansa has hubs in Frankfurt and Munich, Germany. Lufthansa CityLine connects passengers with destinations across Europe to and from these hubs, making more than 300 flights per day. On-time arrivals and departures are essential to customer satisfaction and Lufthansa’s bottom line.
Outside of weather conditions, delays stem from missteps in the tightly choreographed turnaround process. Like most industry players, Lufthansa CityLine relies on manual time stamps to understand when each step of the turnaround process begins and ends and uses this manual time stamp data to gain insight into the adjustments needed for longer timeframes. faster and lighter.
In a pilot phase, the airline partnered with zeroG, a Lufthansa Group Consulting company founded by Lufthansa Systems to accelerate the tangible impact of artificial intelligence in the operational and business processes of airlines around the world. One example is improving lead time management using AI.
ZeroG’s Deep Turnaround Solution Leverages Azure video analyzer, a new offering from Microsoft that combines the functionality of Live Video Analytics and Azure Video Indexer. For Lufthansa, it generates automatic timestamps from video feeds and issues alerts when the rotation script is disabled.
“With this transparency from Deep Turnaround – knowing when the caterer arrives, knowing when the bridge arrives to get off the plane – the airline can lead the process and have much simpler processes than before,” said Manuel van Esch. , principal consultant for zeroG.
For example, when a supply truck arrives later than expected, Deep Turnaround alerts turnaround coordinators and other ground operations personnel. The alert initiates a search for a solution that avoids a delay, such as sending a second tanker to the plane.
Applied AI services
Azure Video Analyzer is one of a handful of Azure Applied AI Services that Microsoft highlighted on Tuesday at Build, the company’s annual developer conference. These services – Azure Video Analyzer, Azure Metrics Advisor, Azure Bot Service, Azure Cognitive Search, Azure Form Recognizer, and Azure Immersive Reader – accelerate the development of scenario-specific AI solutions.
Azure Applied AI services are based on AI models at the heart of Azure AI products and services. This includes Azure Cognitive Services, which offers customizable AI models and tools to create AI solutions that help customers extract meaning from text, integrate speech into applications and services, identify and analyze the content of images and videos, and make decisions.
Customers can also customize these services and extend them with their own custom Azure Machine Learning models to meet their specific business needs.
Customers regularly tell Microsoft that while they see the potential of AI, building solutions is harder than expected, said Eric Boyd, corporate vice president of Microsoft Azure AI in Redmond, Wash.
“The goal with Azure Applied AI Services is to provide a bit more packaging and structure to really accelerate the development of AI solutions for common business processes,” he said.
The Azure Video Analyzer service, for example, combines Azure Cognitive Services computer vision and an automatic captioning model as well as capabilities for integrating closed-circuit video streams and existing video management systems, which facilitates the creation of video analytics solutions for businesses.
Microsoft created the Azure Applied AI Services category to target common business scenarios that Boyd’s Azure AI team has seen customers repeatedly build from scratch. For example, Azure form recognition is based on Optical Character Recognition, a computer vision technology that recognizes text and is the key to many business solutions ranging from reading receipts to extracting data from admission forms.
“To put it in the app they wanted, they had so much more to do,” Boyd said. “It wasn’t just about getting the text, it was about understanding the structure of the document and saying, ‘I have this form that someone has filled out and I want the information to be there in my. database.'”
Azure Form Recognizer builds on the underlying OCR technology with a framework for understanding the entire structure of the document, extracting relevant information, and populating a database.
Rely on Microsoft’s internal AI solutions
Many Azure applied AI services rely on AI tools originally developed for internal products and services, including Azure Metrics Advisor. The tool was born out of the work the developers did to have Microsoft’s Bing search engine detect deviations from normal operations, such as spikes in queries from a country or a sudden drop in ad revenue.
“The research is really pretty predictable in the way it changes from day to day, and so being able to detect these anomalies, we could really jump on issues and fix them faster,” Boyd said. “This anomaly detection service has been deployed in a number of places, such as Power BI. But it’s a development interface and requires you to chain a lot of it. “
Microsoft made the technology publicly available through Anomaly Detector, one of the Azure cognitive services. For Azure Applied AI Services, Microsoft has leveraged the technology that powers Anomaly Detector and adapted it to common solutions for business customers, making it easy to deploy a solution that monitors metrics and, if something goes wrong. , issues an alert and indicates where to look. to solve the problem.
Samsung Electronics has deployed Azure Metrics Advisor in China for anomaly detection and root cause analysis of issues that can lead to failures on the cloud-based hardware and software system that enables 24-hour access to audio and video content posted on the Internet for display. about the company Smart televisions.
AI solution built with Azure Metrics Advisor helps Samsung engineers detect incidents before they affect customers and resolve issues quickly, said Jie Zhang, technical manager of the research and development center at Samsung Electronics (China), which helped with the design and implementation.
The backend development of Azure bot service has followed a similar trajectory to Azure Metrics Advisor, Boyd noted. This service leverages the core speech and language technologies that power Azure cognitive services such as Language Comprehension, QnA Maker, Speech to Text, and Text to Speech to help clients develop conversational assistants. intelligent.
“We combine a number of cognitive services together and package them and make it easier for users to use all of the services,” Boyd said.
More value with AI
Many of the services now available in the Azure Applied AI Services category were previously available as stand-alone Azure Cognitive services such as Azure Form Recognizer and Azure Immersive Reader, which allows developers to implement techniques in their applications that improve reading and writing for people of all ages or abilities.
Other services were individual offerings under Azure AI, such as Azure Bot Services and Azure cognitive search, which allows developers to integrate AI-based search into their apps and websites.
The reorganization, Boyd said, is aimed at making it easier for business customers to find artificial intelligence solutions for common business processes. The new category is expected to grow in the months and years to come as the Azure AI team works with customers in specific industries or sees customers solve the same problem with a combination of Azure AI services.
“We see this category of ‘How do you group and simplify these things?’ really talking to people, realizing that there is tremendous power in what AI can do and saying, “I have to bring this to all areas of my business,” Boyd said.
The ability of AI to generate and make sense of data is driving the zeroG adoption of Azure Video Analyzer for the Deep Turnaround solution powered by Lufthansa CityLine. And that’s just the start of a digital transformation with AI across the Lufthansa Group, according to Xavier Lagardere, the airline’s data director.
“We are not yet a type of business that is driven by real-time data as a matter of course when it comes to making choices, making decisions or even acting on data,” he said. . “There is an exciting road ahead to do a lot more with the huge stacks of data we generate day by day.”
Learn more about Azure Applied AI Services
Read: What is it? Microsoft’s latest breakthrough, now in Azure AI, describes images as well as people
Read: Microsoft’s Responsible Machine Learning Capabilities Build Confidence In AI Systems, Developers Say
Read: Microsoft Services Help Healthcare Organizations Develop and Deploy Virtual Healthcare Assistants
Top image: A Lufthansa CityLine plane is parked at an airport gate while the ground operations team prepares it for the next flight. The airline is piloting AI technology that can help improve the efficiency of the rotation process. Photo courtesy of Lufthansa.
John Roach writes about Microsoft research and innovation. Follow him on Twitter.