Adding Arms to Autonomous Aircraft
In what is perhaps a bizarre twist on both embodiment and robotic research, a new project, backed by considerable funding from the US National Science Foundation is concerned with a practical investigation of the prospects and pitfalls of adding dexterous limbs to UAVs. These arms, complete with grasping digits would unfold from the aircraft to work on nearby structures or aid in combat, or rescue operations.
The researchers involved are from Drexel University's Mechanical Engineering and Mechanics department, and are led by Dr. Paul Oh, a professor in Drexels College of Engineering and head of the department. His work has always been closely tied with extending the possibilities inherent in autonomous systems. Even so, this one is certainly going to prove a challenge.
Whether the robotic limbs would be controlled by the UAV's own AI or radio controlled by operators in the ground, is not even an issue at the moment. Whichever method is chosen and both are equally feasible there is a much more pertinent issue to overcome.
Any sudden movement by these arms, which would either be perched on top of the aircraft, or suspended below it (or, even in both arrangements simultaneously) is going to alter the trajectory of the aircraft, and radically redefine it's aerodynamics. Every movement therefore has to be offset by the aircraft's own thrust control systems in real-time, a feat made even more challenging because of atmospheric variations, and the very real likelihood of close proximity of structures or other hazards.
It would be much more challenging still to compute all the variables necessary to take into account the torque from projectile weaponry held in those dexterous limbs, so the Japanese Anime and Manga's idea of fighter planes with six-shooters in optional hands, is even more far-fetched and unlikely than the rest of this idea.
Dr. Oh envisages the probable capabilities of such UAVs to include then working like cranes on a construction site: picking up girders and pallet loads without any wiring, lifting them up and about the building to where they are required. Bridge and complex structure maintenance is another potential task, with the UAVs able to access where it is difficult to airlift humans to, and proceed to work with just as deft a touch as a human whilst they hover nearby. A third possibility is in rescue UAVs, which can attempt to lift a patient off of a dangerous surface, or under tight time-constraints. Much as the modern air-ambulance services do now, but without risking human crews. Instead the helicopter does the job of both aircraft and crew.
To begin the research in earnest, Dr. Oh and colleagues have constructed a functional mock-up version of how they envisage such an armed UAV operating. The prototype (pictured above) is not mounted to a plane, but rather has its dexterous arms and hands fitted to an adjustable gantry system that is configured to mimic a UAVs lateral and longitudinal movements in free space. In this manner they can gauge how the UAV will respond to different movements of the arms, and how they will directly impact flight performance.
These types of aircraft will advance field service robotics for things like search and rescue and disaster mitigation, Oh said. It could help with infrastructure repair; instead of hoisting someone up to a bridge, these robots might be equipped to fly up to the bridge and start welding.
The three-year $649,999 NSF grant will charge the group with challenges such as developing a a system that would allow the UAV to interact with objects without upsetting its own stability. Oh developed the proposal and he will be joined in the research by Dr. M. Ani Hsieh, Dr. James Tangorra and Dr. Jin Kang. The desirable end result will ideally be an actual prototype UAV capable of extending and using its arms whilst hovering in a controlled fashion above a target task site.
Like all things that fly you want to make sure they dont crash, and as this type of flying robot starts manipulating things in its environment it can often destabilize the vehicle, Oh said. This is a very challenging design problem that nobody else has ever really attacked.