ISG Insights

Automation and AI Concerns – What They Mean and What To Do

Alex Bakker

What is Happening?

As multiple news sources have already pointed out, Hurricane Harvey will likely be a “breakout moment” for drones in commercial applications. Companies like ATT, Farmers Insurance, and Allstate will be deploying drones to facilitate the inspections of infrastructure and buildings to evaluate necessary repairs and, especially, insurance claims. By leveraging on-board cameras and smartphone-based flight control and programming, drones will be used to quickly create composite photographs with enough detail to evaluate (for example) roof and yard damage, without the need for an inspector to climb up onto the roof. As Reuters reports, “Farmers Insurance said a drone could help a claims adjuster process three houses in an hour. Without a drone, only about three houses could be processed in a day.”

Additional considerations factor in as well. Given the scale and magnitude of this storm, there will be many thousands of buildings damaged and in need of insurance inspections – Moody’s has already estimated the storm may have caused between $30 billion and 40 billion in damage. As Kespry – the drone provider behind Farmers Insurance’ drone program – is happy to highlight, drones will likely reduce the injury rates among claims adjusters – a benefit even more important in an area where many buildings have been structurally damaged by the storm.

Why is it Happening?

Last year, the FAA opened the doors to a well-documented process for enabling commercial drone operations. At the time we were very positive because it was a major step forward to an industry that had been operating via FAA exemptions in order to fly at all. With the reduced barriers to entry, many more companies that had been in trial stages, or simply investigating using drones began to create more formal and standardized programs. Finally, emerging platforms, hardware, and acquisitions have created a vibrant ecosystem that is aimed at solving an increasing number of business problems that would have previously been difficult (e.g., roof inspection), expensive (e.g., crop monitoring), or dangerous (e.g., antenna inspection), in a way that is cheaper, faster, and safer.

Market Impact

If hurricane Harvey is a moment for drones to take the spotlight, it will also begin a period where critical developments in the drone industry will determine how long they stay there. Right now, the vast majority of drones are leveraging the combination of GPS and Cameras to generate value. Mapping, Inspection, Crop Analytics, and Insurance claims are all use-cases where the transformative feature was having a camera that could fly. Under current rulemaking, those use cases could also be where the industry stalls, though.

FAA’s Part 107 allows the operation of small (<55lbs) drones, within visual line of sight of the operator, during daylight, not over people, with a one-person-one-drone relationship. While most of these requirements can be waived on a case-by-case basis, additional rulemaking will be necessary for the commercialization of services that fall outside of those categories. For example: while a utility company may gain a waiver to fly beyond visual line of sight (BVLOS) while inspecting powerlines in the woods, they will not be able to leverage that same waiver to fly over powerlines near people’s homes (most likely). Flights over roads, bridges, pipelines, and railroads – collectively called corridor inspections – remain an area with some of the biggest potential for value and the accessibility. Additionally, at the scales at which drone delivery might ever make sense, rulemaking and standards around BVLOS and full autonomy will be prerequisites for real commercialization.

Meanwhile, there are two competing paradigms in the drone industry that tend to go unspoken, but are critical underpinnings in planning and decision making: the dichotomy between drones as sensor platforms, and drones as robots.

From one frame of reference, a drone is nothing more than a cheap, small, and light platform which allows you to fly sensors. It appears that FAA’s Part 107 subscribes to this definition, as does the treatment of drones as an extension of the Internet of Things (IoT). Treating these as “aircraft” and piloting them is also an extension of this viewpoint. They are devices we control, and we can mount any sensor, camera, LIDAR, package, or payload to it, and they become another survey instrument, or dumb vehicle with a simple purpose.

From the other frame of reference, we think of these not as sensor platforms, but as robots – robots that have immense potential to be autonomously interacting with humans on a wide scale, in the physical world, within a very short period of time. (yes, probably before we all have self-driving cars). This is the view taken by Amazon and Google with their package delivery projects, or drone maker DJI with their increasing array of gesture-controlled drones that have on-board subject tracking and image recognition processors, not just as safety measures, but as the primary means of interaction and control with the environment and the people around them.

Today, it is easy to view drones as being limited to the first category – and potentially to even want them to stay there. Concerns about privacy and safety lend themselves toward regulating and controlling. However, given the pace of progress in areas like clean energy generation, computer vision, and AI, it may be more productive and realistic to grapple with the problems in the latter category. Drones may face us not as tools or instruments but as fully autonomous, computer-piloted machines. When that happens, we need to be prepared to use them and interact with them safely, securely, and productively. The aftermath of hurricane Harvey may be drones’ breakout role, but autonomy is what will win the Oscar.

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.