High above in the treetops, flying contraptions dart in and out of the branches, buzzing as they collect vital material. Though they may sound like a swarm of hornets, they are actually more benign, industrious creatures: unmanned aerial vehicles, or drones, inspecting the electrical grid.
These drones, armed with sensors customized by Avitas Systems, a GE Ventures company, collect data for power companies like bees gathering pollen. Patrolling treetops and other hard-to-reach places, they hunt for anything that could damage the grid and cause outages — a mission known as vegetation encroachment. And using digital 3D models, inspectors can even plan out their drones’ flight paths, pinpointing points of interest — perhaps a bird’s nest or branch dangling precariously close to live wires — that require a closer look. “This digital solution is safer, faster and less expensive than sending men out in cherry pickers with automated saws to figure out where to cut back trees,” says Alex Tepper, founder and head of corporate and business development at Avitas Systems.
Avitas Systems has branched out considerably since its launch last year. The Boston-based company started out by focusing its flying, crawling and swimming fleet, artificial intelligence and predictive analytics on large oil and gas companies. For example, Avitas Systems deployed drones to examine some of the industry’s most precarious assets — like flare stacks at oil refineries, which can heat up to hundreds of degrees. To do so, drones must handle searing desert heat as well as Alaska’s frigid air. “Oil refineries are not put in locations with hospitable environments,” Tepper notes.
A Digital Wind farm in Canada utilizes 60+ connected wind turbines from GE Renewable Energy. All the more reason to see it from new angles. #DRONEWEEK http://invent.ge/2hXzbzi
Posted by GE on Monday, October 16, 2017
But the technology found new applications as the year unfolded. In addition to vegetation encroachment throughout other parts of the United States, Europe and the Middle East, aerial drones can also check train tracks for weak rail ties that might cause derailment, and orbit airplane frames to search for dents and scratches, slashing a 90-minute inspection to half an hour. Tepper says that sales teams attend roughly 30 customer meetings per month. “The challenge has been prioritizing customer demand,” he says. “That’s not a bad problem to have.”
Tepper says that one of biggest business drivers for Avitas Systems has been the artificial intelligence (AI) and predictive analytics the company uses to spot flaws and recommend optimal inspection and maintenance schedules. Avitas Systems transmits data to a cloud-based platform running Predix, a software system GE developed for the industrial internet. The platform is one of the few that can digest inspection data collected by different methods — including by humans or from sensors installed around a chemical plant — as well as weather records and other external data.
The system runs the combined data haul through advanced algorithms to detect defects. It then beams the findings to inspectors who can view the results on an online dashboard. “There are many different groups gathering information, but it’s rarely looked at together,” Tepper explains.
Last September, Avitas Systems bulked up its technological muscle by partnering with the supercomputing company NVIDIA. Avitas Systems uses NVIDIA’s AI hardware to enhance its deep learning software and enable the continuous improvement of inspection methods, catching smaller cracks and abrasions with every inspection.
This will come in handy as the company scales its business. In December, for example, Avitas Systems announced that it would team up with Bureau Veritas, a leading inspection company based in France that operates in 140 countries, to train its licensed drone pilots and digitize inspection with Avitas Systems solutions. The company says its customers can anticipate 25 percent faster inspection turnaround, up to 25 percent reduction in inspection costs and 15 percent less downtime for maintenance. That’s a pot of honey well worth the investment.