Compared to a vehicle with a single main rotor, like a helicopter, you’d think a quadcopter would be altogether safer. After all, hasn’t it got multiple possible failure points before it can no longer remain aerial? Unfortunately, that’s not correct. Most drones will struggle to fly unless all four rotors are operational. That’s something that researchers from the Netherlands’ Delft University of Technology have been working to solve.
At the recent International Conference on Intelligent Robots (IROS 2018) in Spain, the team demonstrated a “fault tolerance controller” which allows a quadrotor to continue flying at high speeds, even if a rotor has broken or a motor has failed. No, it doesn’t look quite as pretty doing so, but, crucially, it remains stable. Just as importantly, it is able to do so while retaining forward momentum. This is thanks to some smart math on the part of the Delft investigators, who were able to draw on information such as the drone’s built-in gyroscope and accelerometer to work out how flight should be achieved using the remaining three rotors.
“Imagine that when a quadrotor is delivering an important package over water where strong wind blows. All of sudden one motor malfunctions,” Sihao Sun, a researcher on the project, told Digital Trends. “Normally in this case, the drone will crash into the water together with the package. But with our technology, it is able to continue flying at a considerable speed to fly back to a safe landing place. This could save the package and the drone itself.”
In another use case, it’s possible to imagine how a drone with a damaged rotor could be set to return to its base of operations in a safe manner, without endangering people and property.
The team put the drone (a standard, off-the-shelf Parrot Bebop 2) through its paces in a wind tunnel to simulate forward flight without actual forward movement. The quadrotor reached a top speed of 9 meters per second, roughly half the maximum top speed of a Bebop with four functioning rotors. The team hopes to soon extend this to outdoor flights.
“The next step in this project is the combination of this fault tolerant control technology with real-time fault detection and envelope prediction and protection,” Coen de Visser, another researcher on the team, told us. “An important question that we are hoping to answer with our research is this: How can we make the drone ‘aware’ of its own physical limitations and capabilities after a failure has occurred? Without such an awareness, after a failure, the drone cannot make an informed decision about whether it should conduct an emergency landing, attempt to fly back to base, or even continue the mission with reduced performance.”