Drones in the Enterprise: Data Driven Risk Management

Dr. Earnest Earon is a drone industry veteran – and currently, he’s at insurtech platform REIN, the company behind DroneInsurance.com.  REIN’s expertise is in interpreting risk – and they’re using sophisticated AI and big data science to do it.  But at the DroneDeploy 2019 conference, Earon presented a simple but signficant idea for large enterprise: leveraging large sets of data – sometimes just by looking at it – in order to extract an understanding about risk.

The Data Driven Approach to Risk Analysis

Earon says that the data driven approach is something of a necessity in emerging markets like drones.  “We don’t have hundreds of years of historical ops data in this industry…  and we have very rapidly evolving technology in the space.  These systems have very dynamic manufacturing schema.”

Using data about a lot of flights, combined with things we know about weather, application, and environment, can help stakeholders understand the differences in risk between a flight performed in Florida, for example, vs. one performed North Dakota: and a thermal roof inspection vs. a vertical tower inspection.

Some of the risk points that come out seem unusual, others are common sense.  Platform, for example, makes a difference in risk: and in the drone industry, a thorough evaluation of the top 5 platforms covers most of the drone models being used.  Newer models, unsurprisingly, tend to have fewer crashes.

Other points are surprising.  Flying in the state of Florida, for example, appears t0 be risky: there are a larger number of crashes in Florida than in other locations. Flights in December are more prone to crash than in other months.  And vertical inspections – which still in many cases require manual flight – result in more accidents.  “Any time there is a human in the loop, the chance of a crash goes way up,” says Earon.

From Data to Loss Prevention Strategies

The point of gathering the data, says Earon, is to mitigate the risks.  “We want to take the results and move them down into loss prevention programs.”  While some enterprise operations may wonder why they need to care about the cost of this crash, if it isn’t going to hurt anyone, there are other costs associated. “The problem is operational tempo,” says Earon.  “That’s a significant cost…  It takes the drone out of play, and you still have to send someone out to do the work.”

Earon suggests that companies take a careful look at high risk features and the operations around them, and work with stakeholders to limit the risk.  As an example, if crashes around transmission tower inspections are caused by flying too close to the tower, upgrading sensor quality to enable pilot to fly further away may mitigate risk.

Using a data driven approach in the enterprise can be simple, if not easy: it starts with implementing some program to collect data about all of the flights.  “You need to have some way to centralize and colocate the data so you can use it,” says Earon.  Then look that available data as a whole to identify the risks in your operations. “You can learn a lot just by looking at your own programs.”