Flight testing unmanned systems is more complicated than you may think



By Hannah Lemon, Marketing Publisher for the University of Kansas Edwards Campus

June’s flight of NASA’s Ikhana, a modified MQ-9 Predator B unmanned aircraft, into commercial airspace without a safety chase airplane marked a milestone for sense and avoid development and underscored the challenges of testing unmanned aircraft systems (UAS), according to George Cusimano.

“Programming a UAS to sense and avoid potentially conflicting traffic is a big deal,” said Cusimano – a flight test engineer, United States Air Force veteran and educator with more than 40 years of experience in research, development and testing. “These systems cannot see, they can only sense other traffic. In addition, once the traffic is sensed, the system must decide how to avoid a potential collision in the absence of an onboard human able to perform situation-based analysis. More challenging still is how testers will validate sense and avoid designs without putting the UAS at risk.”

As UASs have become larger and more sophisticated, the flight test mindset has not always evolved as quickly, according to Cusimano. “Modern UASs are complex, highly integrated systems that must be tested with the same rigorous processes as other high-technology aircraft.”

In his courses through the University of Kansas Aerospace Short Course Program, Cusimano shares how flight test engineers, test pilots, test managers, aircraft engineers, aircraft designers and educators can work to make UAS flight test as safe, efficient and effective as possible. He discusses current airworthiness regulations, rules for flying in the national airspace system, the interdependence of software development and model validation, testing challenges and solutions, risk analysis and management, human factors, lessons learned and more through new methodologies, case studies, networking and group discussions.

As he trains flight test professionals, he shares many core principles that apply equally to unmanned systems as they do to manned aircraft, including:

1. Have a well-thought-out plan that details specific test objectives, methodology and data necessary to satisfy the stated objectives.

2. Know what to expect so you can predict what will happen. Never attempt a test point without first knowing how it should turn out. Do not continue to repeat a test point if it is not achieving expected results.

3. When in doubt, don’t. If something doesn’t seem right, STOP and figure it out before proceeding.

“All an unmanned vehicle knows is what has been programmed into them,” Cusimano said. “If something happens – a failure state, an unexpected upset due to wind shear, or a potential collision – the software has to be agile and complex, enough to respond appropriately. This means effectively taking what’s in your head as a human and putting it into a computer without making any mistakes and accounting for all ‘what if’ conditions.” No small task.