Pushing boundaries of drone control for extreme conditions

In order to enhance their operational capabilities, the UAVs must be controllable in all environmental conditions, especially conditions such as in extreme weather or uncertain environments that most aerodynamic vehicles actively avoid, although they will unpredictably encounter a variety of environments.

The presence of a large number of uncertainty and disturbances could disrupt the function of the conventional controllers and lead to significant degradation of performance, causing instability and even damages.

That is where Beibei Ren, an assistant professor in the Department of Mechanical Engineering at Texas Tech University, and her research group are focusing their efforts, building on-board intelligence into the flight control system to accommodate unknown perturbations and enable adaptation to extreme flight conditions.

"We want to push the boundaries of the control theory to maximise the impact of control for different systems," said Ren, who heads the Dynamic Intelligent Systems, Control and Optimisation (DISCO) research group.

"For UAVs, we are particularly interested in developing advanced control strategies that will allow the deployment of the UAVs under extreme and uncertain environmental conditions to accomplish complex tasks."

Such challenging conditions include anything from flying the drone into close proximity to a wind turbine to measure the health of the turbine to landing drones on ships, where it is anything but a smooth process given wind and sea currents as well as the motions of the ship itself.

Ren, who possesses expertise in nonlinear control theory and stability analysis, has conducted research in modeling and control of UAVs. She is the lead author of the research monograph entitled "Modeling, Control and Coordination of Helicopter Systems."

Recently, Ren and her collaborators have developed a novel uncertainty and disturbance estimator (UDE)-based robust control for UAVs to enhance their operational capabilities in extreme environments. The basic idea of UDE-based robust control is to quickly estimate and compensate for uncertainties and disturbances, which significantly improves system stability and robustness.

The preliminary results in both simulation and experimental studies have shown the UDE-based robust control is easy to implement while bringing better performance than other existing robust control methods.

"What we are developing are drones with everything on board," Ren said. "We're developing drones with the control intelligence system on board, which means the drone is ready to fly anywhere, anytime and can do different tasks".

"Suppose you have a different model, a different system and your control system algorithm needs to be changed. Sometimes you have differences that make the system parameters or properties different, so you have to design a vast control system to fly on those platforms safely and accurately. That is what we are working on from a research perspective."

One possible application where UAVs can be useful, Ren said, is in infrastructure health monitoring. That could involve tasks such as oil and gas pipeline monitoring, checking the structural integrity of bridges and roads or mobile highway traffic monitoring and coordination.

One part of particular interest in West Texas is using drones to monitor wind turbines, especially on the large wind farms of the South Plains. Instead of having one person out checking the health of wind turbines, a fleet of drones could get the job done much more efficiently.

But there are issues with flying drones so close to the turbines, which can cost upward of $2 million. The most pressing of those issues involves proximity. The accuracy of GPS technology -- or how close a UAV can fly to a turbine, is roughly 2-5 meters.

However, that is not close enough to take accurate readings using instrumentation to determine the integrity of the turbine blades. So Ren's team is working to develop control systems that would allow the UAV to fly closer to the wind turbine, taking into account the wind currents created by the giant blades as well as other factors.

"The Department of Energy has recently mentioned they are interested in having drones for infrastructure health monitoring, but they have big concerns," Ren said. "If the drone control system is not done well, that might happen where the drone would fly into the turbine blades. That requires very accurate positioning."

That accuracy, Ren said, depends on the sensors and controllers on board the drone, and that is the focus of Ren's research group, developing the on-board intelligence that allows drones to adapt to all possible variables associated with a particular place or region.

Another possible application of drones is for at-sea search and rescue. However, landing a drone aboard a ship at sea can also present a unique challenge.

Shipboard landing requires precise relative navigation and ability to maneuver in highly turbulent ship air wakes and landing on pitching and rolling decks at high seas. Plus, at sea, wind gusts may affect the operation of the drone, and in the middle of the ocean, GPS technology or visual acuity from cameras won't always be available.

Determining if the control systems are an improvement, however, relies on the interpretation of data collected when they are tested. To that end, determining the data's accuracy and usefulness requires experts in various fields of study.

What Ren and her fellow researchers can determine is if the drone's control system operated effectively to allow for that data to be collected in the best possible way. They also can use past data to improve the drone's stability and improve its trajectory.

"Accuracy depends on the sensors, and that control scheme is what our experience is in," Ren said. "That's called on-board intelligence. We can work with those experts to do real-time work interactions and collaborations. But our goal is to improve the flight and stability and to provide data collection as well."

One improvement to drones that could seem fairly obvious but that has yet to be solved is battery life. As it stands now, the average flight time for drones such as the quad-copter variety Ren uses is about 10-30 minutes. If it carries more sensor equipment, that extra weight will drain the batteries even quicker.

Though there has been extensive research on drones, batteries remain restrictive. A solar-powered drone could be one idea, but Ren said the conversion rate of the solar panels is not that good, leading to ineffectiveness.

The solution could be not in the battery itself but in establishing wireless charging stations. Wireless charging could be performed regardless of whether the drone is flying, hovering or landing on a charging station, Ren said.

Plus, having numerous charging stations in a certain area would allow drones to not only recharge quickly but increase their use, which could be particularly helpful in things such as infrastructure health monitoring.