TORONTO – Statistically, flying is the safest way to travel. We don't worry about airplanes dropping from the sky. But drones are another thing altogether.
If a drone runs into mechanical problems, there's no Chesley "Sully" Sullenberger to land it it the Hudson River. To keep unmanned aerial vehicles (UAVs) from landing on our heads, NASA is trying to make them smarter.
Dubbed Safe2Ditch, the technology is aimed at allowing drones to continuously run self-diagnostics during flight to anticipate problems. If something goes wrong, the system could make changes to how the drone is flying and estimate how much longer it could stay in the air.
Since a drone with mechanical problems would need to set down quickly, Safe2Ditch would immediately begin to search its database for safe landing locations and autonomously land at the closest spot. Safe landing options would include fields, parking lots or parks, said Lou Glaab, assistant branch head for the Aeronautics Systems Engineering Branch at the NASA Langley Research Center. Worst case scenario, a drone might have to land in a dense forest to avoid people, but the goal is to keep avoid damaging the drone in an emergency landing.
In a telephone interview with EE Times, Glaab said military drones have a very different concept of operations, with different requirements for what they can do. “We can't operate military drones in our airspace,” he said.
However, there plenty of accident reports involving drones near military facilities, and often one of their main tasks is to infiltrate another country's airspace. “There's a cost component in commercial UAVs that aren't in military ones,” said Glaab.
The goal of Safe2Ditch is to cost-effectively give drones the intelligence to crash land if there's a problem without putthing people or critical infrastructure at risk.
Commercial UAVs must fly close to the general public, so reliability is critical, Glaab said. That where's the idea for Safe2Ditch came from.
The challenge is that it's not cost-effective to use the same expensive systems that manned aircraft employ to make them reliable, said Glaab. “Commercial drones can't be half a million dollars each,” he said.
Commercial UAVs also have to be light weight, so to fulfill all of these requirements, part of the effort around Safe2Ditch has been to repurpose technology that's already available. For example, integrated vehicle health management is already an area of focus in the commercial transportation sector, and terrain databases accurate enough for Safe2Ditch's purposes can already be found in applications such as Google maps, said Glaab.
For a Safe2Ditch-equipped drone to accurately gauge a safe landing spot, it needs sensor and camera technology adequate to the task of identifying a nice little field. Glaab said machine vision technology used in supersonic jets is being leveraged which was developed to address an expensive problem seen in the now retired Concord: the nose had to be mechanically lowered so the pilots could see during takeoff and landing — a mechanical feat that was quite expensive. This is being is replaced with heads up display technology that could be adapted for drones, he said.
A Safe2Ditch feasibility prototype.
From a compute perspective, Safe2Ditch is also leveraging what's already been developed, said Glaab. However, a great deal of the effort is around software that makes assumptions around what will be on the average commercial drone, such as the camera. To address the weight and cost issues, it makes more sense to integrate with the existing components, rather than stipulate new ones, Glaab said. The expectation is this approach will consume less power and be less expensive.
“From a NASA perspective, it's not in the game of competing with industry," Glaab said. "It's a lot cheaper to buy components.”
Discussions have taken place with UAV operators, said Glaab, but ultimately it's those who make drones, not buy and use them, who will need to provide feedback and guidance. The market is used to certain price points, so one goal of Safe2Ditch is make sure it meets the requirements and budgets of a small carrier.
But Safe2Ditch is not just about making courier drones safer as they deliver packages, said Glaab. Reducing the risk of UAV flight in populated areas will benefit people in other ways, such as disaster relief. “There a great applications and use cases just by getting a simple camera in air, not just the delivery of shoes from Amazon,” Glaab said.
—Gary Hilson is a general contributing editor with a focus on memory and flash technologies for EE Times.