Advancing Commercial Autonomous Drones – IBM Business Operations Blog

From Amazon announced drone delivery testing almost a decade ago, commercial drones continued to fly in various industries. Organizations around the world deploy commercial drones for deliveries or to collect video or images with a built-in camera. Drone inspections are already popular in several industries, but the commercial use of drones got a boost with recent message that the UK will become the world’s largest automated superhighway within two years.

What are autonomous drones?

The European Union’s Aviation Safety Agency defines an autonomous drone as “capable of flying safely without the intervention of a pilot […] with the help of artificial intelligence, which allows it to deal with any unforeseen and unpredictable emergency situations.’

If drones can use artificial intelligence (AI) to determine when to take off, which direction to fly, how to deal with external factors and how to return after the mission is complete, there will be less need for pilots or drone operators.

Why are unmanned aerial systems traffic management and ‘beyond line of sight’ critical to the future of commercial drones?

Many countries are drafting regulatory frameworks for managing low-altitude traffic to accommodate the future of drones. This framework will cover roles, responsibilities and protocols for sharing data as part of drone operations. In the US, federal agencies are creating an Unmanned Aircraft Traffic Management (UTM) system. In the UK, the Civil Aviation Authority (CAA) is working towards something similar.

Just like road vehicles, drone identification is part of the UTM requirement, and the US FAA already requires all drones to be identified. There are also plans to include Remote ID for drones, which will provide identification and location information that others can obtain.

In the US, a drone operator is required by law to have visual line of sight (LOS) to the drone. To enable large-scale commercial use of drones such as the UK’s superhighway, the regulatory framework must allow ‘beyond visual line of sight’ (BVLOS) piloting. Some countries with large amounts of remote locations, such as Iceland, Norway and Sweden, have already activated BVLOS as a means of helping isolated communities.

How can a drone in a box help?

Frost & Sullivan defined a drone in a box in a 2018 report on drone delivery: “Sensor, communication, hardware and software technologies have advanced to the point where innovative companies can offer semi- or fully autonomous vehicles that can be automatically launched and returned to base stations or enclosures. These solutions are often called “drones in a box” because structures are required to recharge, protect, or recharge and protect drones between mission stages or between different missions.“

Today, many companies are investing in a drone in a box as a core component of future industrial drone operations. At least two other features will become part of this component when 5G becomes the connectivity infrastructure:

1. Transfer drone imagery from the base station to the processing organization to enable near-real-time decision making.
2. Support for edge computing so the drone can be guided based on what it sees during inspection.

What supports will maximize the benefits of commercial drone inspections?

When implementing drone surveillance technology in an enterprise, it is vital to create a plan that maximizes efficiency. By combining drone images and videos with these seven technologies, businesses can automate workflows and improve the productivity of their business operations.

1. Storage for images

Drones capture high definition videos of the infrastructure or assets to be monitored during a drone inspection. In general, a 4K resolution video shot at 30 frames per second needs about 760 MB of storage space for each recorded minute. Recording drone footage for even just a few days can add up to terabytes of storage space. Because of this, businesses have realized that cloud storage is a cost-effective way to store and archive footage for later analysis.

2. Stitching images

Linking videos or images allows companies to see the full structure of an asset, rather than spending time and money monitoring footage for changes. This is especially useful on large structures such as bridges or construction sites. This effective tool helps managers identify problems and monitor the pace and progress of solutions.

3. Other relevant data sets to support the analysis

Initially, analysts can simply compare the change in the asset over time. When additional aspects of the data are included, planners can achieve richer interpretation, analysis and pattern discovery. For example, when data on rust rates, types of rust, types of structural rust damage, and types of weather patterns are used with drone images of a bridge, it helps predict areas of potential structural damage versus surface changes.

Urban planners can quickly complete planning activities when existing data on nearby topography, buildings, roads and infrastructure are used in conjunction with drone imagery of a specific area. Likewise, using weather data such as temperature, wind direction, rain potential and past wildfire data can help experts monitor and identify change more quickly than with drone imagery alone.

4. Finding patterns with people

A bridge inspector who has spent 20 years on the job is able to look at a particular crack or split in concrete and immediately tell if it is cause for concern or not. The expert surveyor considers depth, color, location and other factors to make this assessment. Human experience and knowledge help identify patterns to create appropriate data sets that can train computers.

5. Computer vision

Computer vision trains AI to identify the same patterns that an expert inspector would see. For example, by training computers to identify images of various concrete cracks, AI can automatically monitor bridge images to locate defects. This complementary drone solution eliminates the need for people to go through hundreds of hours of drone footage.

6. Rules and Decision Making

Once humans identify a set of patterns in images and teach AI to do the same, organizations can create business rules. For example, if a certain type of structural defect is found on the roof of a house, run the drone inspection again after x months to see if there has been a change. In a more critical scenario, such as if the construction plan and actual data are out of sync, various departments will be prompted to act immediately.

7. Digital twins

Drone mapping of a building or set of structures, such as a telecommunications tower, can help create a digital twin. This digital twin can then help companies understand how the physical asset is performing based on real-world data. For example, with a digital twin, organizations can study and predict the ways in which a hurricane may affect the position of mobile antennas on a telecommunications tower. These forecasts can inform engineers of individual towers that will need maintenance before such an event. Moving from reactive to predictive maintenance can save organizations from downtime and keep business operations more efficient.

Evaluating the use of drones in your organization

Automating inspections and reducing manual efforts with drones will drive significant growth in the coming years. The global drone services market is is expected to grow from US$ 9.56 billion in 2021 to US$ 134.89 billion by 2028 at a CAGR of 45.9% during the forecast period 2021-2028. As this market grows, these industries will require more specialized services with drones.