Over the last decade, drones have advanced from multimillion-dollar military machines to being small and affordable enough to be used recreationally and commercially. Now they are revolutionizing a wide range of industries and saving time, money, and potentially lives in the process.
Until recently, drone technology relied heavily on human controllers and acted as carriers for cameras and instruments which would then be monitored manually. But, in news that will strike fear into the heart of any Terminator movie conspiracists, experts are now merging drones with artificial intelligence (AI) systems with incredible results.
My firm recently carried out two technology scouting projects that had us assessing many of the players in this field. We found that drones fitted with AI technology are being used to disrupt a number of industries and undertaking a range of tasks with minimal human input.
Machine vision AI in industrial inspection
In the multibillion-dollar energy industry, tall, hard-to-access structures such as oil rigs, wind turbines, and storage tanks need to be regularly assessed for structural damages for safety and insurance purposes.
This role has traditionally been the domain of fearless rope access teams, who would manually check structures while dangling hundreds of meters above the ground, often in perilous weather conditions, which has led to multiple accidents and even fatalities worldwide.
According to WindSpect, a company that specializes in providing technology for structural assessments of wind turbines, aside from being extremely dangerous, these manual checks are time consuming, expensive, and inaccurate. It can take four to six hours to inspect all three blades and to create a report specifying damage discovered and suggestions for upkeep and repair.
In recent years, experts have used video and photo footage gathered by drones equipped with high quality camera to make assessments quicker and safer, but the photo and videos gathered still had to be interpreted by a human. Now, new AI tech is entering the market that uses machine vision AI to assess structures, effectively removing the need for human controllers to assess footage and greatly reducing the overall time required for inspection.
Machine vision AI effectively allows a computer to see, using video cameras, analog-to-digital conversion (ADC), and digital signal processing (DSP). The farmed data is then sent to a computer or robot controller to be assessed for anomalies. The benefits of this new tech are numerous.
The drone’s machine vision software allows the images and videos to be analyzed automatically to detect and classify objects and anomalies such as hidden cracks, foreign bodies, depositions of water, loose adhesive joints, and uneven coating.
Wind power is booming, and the potential for such technologies is enormous. According to Navigant Research, the “cumulative global revenue for wind turbine UAV sales and inspection services is expected to reach nearly $6 billion by 2024.”
Object recognition using deep learning on construction sites
The construction and construction aggregates industries are using drones for a number of tasks, from mapping out sites and comparing with architectural blueprints to record progress to keeping inventories of multimillion-dollar construction vehicles and building material supplies.
Keeping up to date with activity across sprawling, complex construction sites is particularly challenging because there are so many moving parts and so many people using equipment for different task. A report by the National Research Council of the National Academies argues that construction lags behind other industries such as manufacturing in terms of productivity and blames the situation on problems with planning, coordination, and communication between on-site teams and those higher up in the chain of command.
Using deep learning AI technology and drones, Silicon Valley-based commercial drone company Kespry is using NVIDIA Jetson TX1 AI technology to offer construction companies a way to keep track of their expensive kit and allow them to remotely manage multiple work sites at any one time.
NVIDIA claims that its credit card sized Jetson TX1 is the first embedded computer designed to process deep neural networks and that it allows the drones to autonomously identify and classify construction vehicles, building materials, and other structures.
While a human element is still important to review inventories and keep track of workers, the new tech will allow site managers to effectively manage a number of sites, in various locations, from the comfort of their computer screens.
Deep learning AI for mountain rescue and police work
In the past, police forces around the world have employed drones in many inventive ways, including monitoring sporting events, doing fly-overs of potentially dangerous situations, and controlling crowds with pepper spray. Now, advances in machine or deep learning AI are opening up new opportunities for everyone from police forces to mountain rescue teams.
For example, the Croatian Mountain Rescue Service has teamed up with the Faculty of Electrical Engineering, Mechanical Engineering, and Naval Architecture to develop a system to automatically analyze aerial photographs taken during a drone search to find missing people.
These drones can extensively scan an area of one square kilometer in just over 15 minutes, saving hundreds of photos, which are then analyzed using algorithms which flag possible finds for human controllers.
Experts have raised concerns that this technology would not be as effective in urban areas, since the large numbers of people and vehicles would distract the robots. However Neurala has created real-time people-locating AI that, it argues, will be perfect for cities and countryside alike.
Neurala’s software-only solution can quickly find people in an image and differentiate between people in a crowd. The computers to be ‘taught’ the face or profiles of a person, allowing drones to track these people from a distance, relaying information and video back to human controllers.
This new marriage of drone and AI is paving the path for revolutionary changes over a number of industries, which will save time, money, and lives in the long run. Critics claim that smart drones are a risk to our privacy and/or that robots are taking jobs out of human hands, but when there are gale force winds and poor visibility, I know who I would rather have checking the top of a 300 foot high wind turbine.
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.