By Linda Fresques
Research focuses on enabling flight of unmanned aerial helicopters
”Robots that fly” is how Ou Ma, professor of mechanical and aerospace engineering at New Mexico State University, describes the unmanned aerial vehicles that he and his students have been working with for the past two years. These sophisticated vehicles could have wide-ranging applications – from military surveillance to crop monitoring to perhaps, someday, pizza delivery.
“UAVs have tremendous potential for military, industrial and civilian applications,” Ma said. “They could be used to gather surveillance information for police. They could be used to survey crops’ conditions or natural resources. If a national disaster occurred, they could be sent in quickly before humans can safely assess the risky areas. Maybe even one day they could be used to deliver pizza in congested urban areas.”
Ma runs a UAV laboratory where he and his students are currently conducting three research projects that involve unmanned aerial helicopters. The projects – all geared toward enabling the helicopters to fly without human assistance – are funded through NMSU’s Physical Science Laboratory by the U.S. Air Force.
Ph.D. student Khaled Hatamleh is developing a system identification method for identifying the mathematical model of a “desktop helicopter” that is mounted on a test rig directly connected to and controlled by a computer. Data about the position, acceleration, velocity and other parameters, detected from several sensors onboard the model helicopter, are communicated to the computer in real-time, as the model vehicle moves, to identify or update its mathematical model.
“We need a good math model for proper control of an autonomous UAV – each time we record data the math model becomes more accurate and will result in better control of the autonomous vehicle,” said Hatamleh.
Complicating the problem is the fact that UAV manufacturers do not usually provide math models of the vehicles they sell and the model parameters may change during a flight. As fuel in the tank is depleted, mass and center of gravity change. Weather conditions also will change some model parameters.
Currently, Hatamleh is calibrating the sensors of the helicopter that he relies on for system identification to ensure they are taking correct and consistent measurements.
Another goal of this research is to supplement the traditional navigation sensors, such as GPS, rate gyros, compasses and sonar sensors, with small cameras to enhance situational awareness and to enable visual servo control because visual data has much more information about the surrounding environment than any other sensors. For example, data from a camera could use perspective to determine relative position. If you look at an object on a wall, it changes shape and size depending on the angle and distance from which you view it, explains Hatamleh. The camera communicates this kind of information to a micro-controller that makes decisions and guides the UAV’s movement.
Using another type of controller, Ph.D. student Jesse McAvoy is working with a small radio-controlled helicopter. The Thunder Raptor 90 is equipped with an autopilot system that uses electronics that can be programmed to control flight. The goal is to let the helicopter take off, fly to a prescribed location and land on its own. A GPS system determines the UAV’s current location; an initial measurement unit tells its attitude; an electric compass gives the flight direction; and a sonar sensor measures the distance from the landing or taking off spot. A visual sensor and control system will be added.
One of the challenges with the development of autopilot capability with this “toy” RC helicopter (i.e., not designed for research or professional applications) is to deal with the severe vibration problem. The strong vibration caused by the gas engine and the mechanical structure prevents the delicate sensors from working properly. To deal with this problem, the sensors and electric unit must be isolated from the vibration. In the research project for his master’s degree, McAvoy developed a novel vibration isolation fixture that successfully isolated more than 98 percent of the vibration. As a result, the sensor system is now working well with this non-research helicopter for the purpose of research.
In a third project, master’s student Carlos Ortega is working with tiny, bird-sized micro air vehicles. Because of their small size and power, MAVs usually have little or no payload capability. To avoid crashing, a UAV is usually constrained to a mechanical arm for testing its flight control system, especially in the early test stage. When an MAV is attached to a test arm, it is not capable of flying anymore because the flying MAV does not have enough lifting capability to carry the heavy test arm. Ortega solved the problem, designing a gravity-balanced arm that can compensate all the weight of the arm, so that the MAV can easily fly while attached to it, like carrying a weightless load. With such a device, an MAV can fly freely while different types of controllers are tested.
“MAVs are just small flying robots – like artificial birds,” Ma said. “These robots are still not very practical because many enabling technologies are still immature. In particular, these vehicles are so tiny that they are incapable of carrying a payload and they would not fly with required sensors and actuators attached. A standard way for early testing an MAV is to attach them to a mechanical arm with multiple degrees of freedom, but you have to somehow compensate for the gravitational pull and mass of the arm.”
The group will use the mechanical test arm to help them learn to manually pilot the helicopters. McAvoy is the only student capable of manually flying the UAVs. Two other students are currently learning the pilot skills.
“It’s been his hobby since he was a child,” Ma said. “He’s the only one I trust to fly our various models (priced from $50 to $200,000). Some of them are too expensive to afford a crash or damage.
“This research effort is necessary for test driving autonomous flight controllers without a risk of damaging expensive equipment and losing time-consuming work,” Ma said.
The group hopes to one day apply their work to the new Yamaha RMAX helicopter that was donated to NMSU by Northrop Grumman in 2009. McAvoy has already made a manually controlled test flight of the $200,000 helicopter.
Ma and his students have been working in robotics research for many years. Their research work is mainly for aerospace and biomedical applications. They became involved in sponsored UAV research only two years ago, thanks to funding support from PSL and the U.S. Air Force. Ph.D. students Hatamleh, McAvoy, Pu Xie and Qi Lu; master’s student Ortega; and undergraduate students Gerardo Martinez, Joshua Choo, Brandi Herrera and Bethany Davis, are working on the UAV projects.
Ma’s group is actively mentoring undergraduate capstone design projects and undergraduate research assistance projects. From 2006-2008, a senior capstone project developed an RC-controlled blimp with funds donated by The Boeing Corp. and PSL. His students have been involved in the NMSU nanosatellite projects sponsored by the Air Force Office of Scientific Research and the C-9 Microgravity Flight Projects sponsored by NASA and the New Mexico Space Grant Consortium. The group is actively working with the NMSU Alliance for Minority Participation Program to recruit and train prospective minority students. The UAV lab and the student team have hosted numerous lab tours and demonstrations for K-12 students in the past two years in support of the university’s outreach program.