isMobile pushes AI in Field Service Management forward. This is a fact and a part of our long term vision. But how is this done and what does it take to be at the front of new technology? Today, we want to introduce you to our data scientist and machine learning engineer making all this happen at isMobile.
So, Joel, tell us a little bit about yourself
Yes, I’m a bit of a combination of a data scientist and machine learning engineer here at isMobile. I have a master’s degree in machine learning from the Royal Institute of Technology in Stockholm and have previously worked at the companies Ericsson, Saab and Predge. I have done research in anomaly detection and time series analysis, but my primary focus here at isMobile has been various forms of image analyses to increase field worker productivity.
You have been here at isMobile since about one year now. What attracted you to isMobile?
I sat down and had a long talk with our CEO Mikael, and he explained in what direction he wanted to take AI at isMobile, what business opportunities there are and how we can make a real impact on our customers. That’s always been one of my primary drivers, producing useful things and having happy customers.
What interesting AI projects have you participated in before you started at isMobile?
As I said I have worked at Saab with anomaly detection in surveillance data. Having an automated system that continuously monitors the behaviour of various parties is an enormous help as the amount of data that is available today is certain to overwhelm a human operator or analyst. Without automatic anomaly detection important happenings will be missed. At Predge I worked with predicting when maintenance was needed to avoid failures and expensive stops in railway and industry applications.
What do you think are the challenges of AI in general in today?
AI has been beating humans at specific tasks for over 20 years now, but they typically generalize poorly. A specialized chess AI typically can’t play poker. Making AI that can perform several tasks, or creating smarter ways to combine several AI that specialize in different sub-tasks, that is an area that is not very well explored today, and will make a huge difference in real world applications.
Are there any typical areas AI is suitable for use in practice?
There is some low-hanging fruit. If you have a task that is repeated many times by a human but is so complicated that it is hard to write a conventional computer program to solve it, it is probably a good candidate for an AI.
You have researched and developed an AI application for meter reading, called smartReader. Can you briefly describe what smartReader is and how AI comes into the picture?
Our customers take lots of pictures of electricity meters. smartReader is an application for Android devices that takes those pictures and automatically extracts information that the field worker would otherwise have to enter manually, at the risk of mistyping. As a bonus, smartReader almost certainly does this faster and makes it less of a chore.
What practical benefit does smartReader have for the customers who use it?
My hope is that smartReader will enable our customers to do better work in less time, and that the end user will enjoy using it.
In addition to smartReader, what other interesting research projects and applications have you worked on?
I’ve worked on data analysis for one of our customers, researching what causes the success or failure of a meter exchange project. A good first application of the findings could be to use them to optimize bidding on new projects, and in the longer run they can be used as input to improvement to their way of working.
The reason of innovation and new technology is to add more value to the users. How would you describe the customer value you have added to the Field Service Management arena?
The value of the project is that the readings can be made more efficiently and more precise. I have developed a smart readout app with object detection and character recognition making the life easier for the fieldworkers. It reduces time spent on ocular reading and writing meter data by hand during service or exchange of meters. It also increases quality by minimizing errors and required user input by intuitive UI and smart layout with smart display detection and formatting aids.
What outcome are you most proud of?
I’m most proud of what I’ve achieved with smartReader. I have tested the application quite extensively and it “just works”. It should really make a difference to the field workers of our customers.
What challenges have you been facing?
The problems I have been working on are active research areas that are by no means “solved”, but there is a big machine learning community out there and we try to help each other. Machine learning in general and deep learning in particular are fields that are moving very fast, so you really need to work to stay up to date with the latest findings.
What makes the hours fly away at work, what catches your complete attention?
Whenever I have a problem and I think I know what to do to solve it. The nature of the problem doesn’t matter much, I just want to crack that nut.
How would you describe the kind of organisation and environment it takes to support this kind of innovative work?
It takes long-term commitment from the organization, and the environment should be one where curiosity and learning are fundamental parts.
How do you think AI will affect our everyday life in the future?
We already use AI every day of our lives, either actively or passively. Even if you go to your cabin in the woods for a week, lock your phone in a closet, and live off the grid with no electricity, various AI will still work tirelessly towards your disadvantage or benefit. A good example is spam filters, when you get back from your week off the grid, they are the only reason your inbox isn’t filled with a ridiculous amount of spam e-mail. In the future AI will be an increasingly large part of our everyday lives, but it’s happening gradually so you won’t necessarily notice unless you really think about it.
A lot of people are afraid of new technology and the big data that comes along, especially when it comes to information security? What do you say about that?
The data gives us an opportunity to boost productivity and improve our quality of life, to everyone’s benefit, but we shouldn’t use the data recklessly. Information security has to be taken seriously and potential consequences of misuse of AI implementations should be considered thoroughly, just as with any big leap in technology.
If you were to predict the future and adding a little bit of wishful thinking. How will AI have changed the work in Field Service Management?
I think we will gradually introduce more tools like smartReader that help the field workers, but also develop products aimed more at back office work.
At last, what is your next project, what will you look at next?
We are already looking at the possibility of expanding the scope of smartReader, allowing it to capture more kinds of data. We will build infrastructure or pipelines for customer-specific adaptations and we are looking at different ways to deploy the AI. In parallel, we are experimenting with AI in end-customer communications, data analysis, and back office automation.
For further information, please visit www.ismobile.com
and find our contacts at www.ismobile.com/contacts