In April of 2009, Ted Ciamillo and I decided to design and build a full-sized ornithopter that would enable a human to fly like a bird. Project Falcon is our cooperative effort. This web page shows my thinking on the matter so far. Whatever it turns out to look like, Ted Ciamillo intends to build the machine and fly it.
We believe that, with the advent of carbon fiber composites, expanded foam structures, exotic alloys, computerized electronic control systems and modern, compact power sources, the dream of flying like a bird might actually be possible now. We aim to begin construction sometime in 2011.
The Falcon
The Falcon is a light-weight, strap-on, exoskeleton aircraft coupled to you through a computer controlled interface. It transforms your arms and hands into bird wings. You both wear it and lie on it, stroking your way through the air almost as effortlessly as you desire. The wing movements are driven by hydraulic pistons pressurized by a simple, compact power source. This particular version (the 8a) is to be powered by hydrogen peroxide. Other power sources are being considered, including lithium polymer battery packs coupled to solar arrays.
The Falcon 8a, has the wings and hydraulic actuator unit mounted behind your shoulders. You wiggle into the exoskeleton cage, sticking your arms out through the side openings and strap the whole thing down like a backpack. Then, boot up the system, grab hold of the hand grips and you're ready to fly. As long as you hold on, the exoskeleton tracks your arm and hand movements and responds with corresponding movements of the wings. Your arms and hands feel the forces on the wings. If you want to look through binoculars, take some photos or just rest your arms, no problem. Letting go of the hand grips automatically puts the craft into its default glide mode configuration.
Learning to fly this machine will be a matter of learning by feel what works and what doesn't, just the way you learned to walk or swim. You will be appropriately tethered in a wind stream or trained on a simulator, or both.
Specifications
- Wingspan: 42 feet
- Wing area: approximately 150 square feet
- Weight: 50-60 pounds, including power and control systems
- Power and control system: hydrogen peroxide-pressurized hydraulic actuators controlled by a negative feedback, computerized, electronic control system
- Controls: Each wing is independently controlled through right and left haptic interfaces
- Flapping limits: plus or minus 30 degrees (restricted to the power-stroke plane)
- Pitch/roll: rotation of the entire wing about the pitch-control axis
- Wing folding: pilot powered
- Joint flexion: the elbow joints flex in the horizontal plane (so the wings can fold up). The wrist joints are more complex. They flex vertically downward (during the back-swing) and have radial deviation in the horizontal plane that is linked mechanically to rotations at the elbow joints (the wrist and elbow joint rotate in unison as the wing folds up and stretches out).
Flapping Flight Made Simple
Edward Tufte's video, taken at 300 frames-per-second, of geese taking flight reveals that the flapping flight of birds is less complicated than previously thought. Birds employ a relatively simple scheme to accomplish what appears to be a complex set of motions..
The power-stroke of the flapping cycle is simple and straight forward. The wings sweep downward and forward while stretched out flat with the joints locked.
During the backstroke, as the humeri are pulled back and upward, a bird simply relaxes its wrist joints and allow the outer sections of the wings (its hands) to fall limp. Simultaneously, the bird folds its wings, pulling them inward. The relaxed hand-sections, caught up in the wind stream, are quickly lifted and thrown backward requiring little or no effort on the part of the bird.
Further simplifying the process, a bird's forearm bones (the radius and ulna) function as a coupling mechanism between the hand and humerus. As the wing folds, this parallelogram arrangement forces the hand to rotate in the horizontal plane in unison with rotation at the elbow joint. This keeps the hand aligned more or less parallel to the humerus.
As shown below, the Falcon wing includes such a parallelogram scheme in its forearm section. The illustration shows the wing skeletal structure in three stages of extension. Note that the hand section stays parallel to the humerus section in all three positions.
Unlike in a bird, the humerus section in the Falcon wing is also a parallelogram mechanism. Its purpose is to prevent any forward or backward motion of the wing as a whole. It allows the wing to fold but not sweep forward or backward. As will be explained below, this makes the power-stroke function much simpler and provides for a clean separation between the three control functions.
Flapping motions of the wing (the power-stroke and backstroke) are confined to a single plane called the power-stroke plane (illustrated below), determined by the tilt of the shoulder joint axis. The power-stroke plane is both down and forward (as it is with birds).
This arrangement allows for a mechanically simple power-stroke in which the stroke angle can be adjusted for optimal efficiency and the entire flapping function is cleanly separated from the wing-folding and pitch-control functions. More importantly, it allows the entire flapping sequence to be powered by only one hydraulic piston per wing.
A bird's hand section is equivalent to the business end of a boat's oar, driving the bird forward as well as upward. The Falcon's hand section, as in a bird wing, will be designed to twist during a power-stroke to tilt the lift vector more forward than it is for the inner two wing sections. In the bird wing and the Falcon's wing, the inner two sections maintain a relatively fixed angle of attack to the wind and so provide varying amounts of lift during all phases of the flapping cycle.
All of these factors add up to a simplified flapping cycle for the Falcon that insures a clean separation between the various power and control functions.
Wing folding
Birds fold and extend their wings during flight for many reasons. They fold them to gain airspeed, to dump excess forces, to maintain headway in a stiff wind and, as mentioned, they fold them during the backstroke portion of the flapping cycle.
Since the lift and drag forces acting on the wings are generally perpendicular to the force needed to fold them up, our ornithopter wings are folded up using only the pilot's arm muscles.
The Falcon wings are tension loaded to maintain full extension. Countering this are the tension forces in the stretched wing-covering membranes that try to fold the wings up. These opposing forces will be engineered to sufficiently balance each other to allow the wing folding process to be done by arm-power alone. You simply pull the hand grips inward to fold the wings and relax your arms to let the wings spring back to their extended positions
Cable-gears
The wing sections are coupled to each other by pulley-like devices I call cable-gears that are built into the ends of the long parallelogram links. The devices function just like regular gears to keep the folding angles equal but are lighter in weight. The left illustration below shows the cable-gear arrangement of the right-wing elbow joint. The red cable wraps around the gear on the forearm beam and crosses over to the trailing-link gear of the humerus. The black cable similarly connects the other two gears. The cables cross in a figure-eight fashion at the center of the joint. In the wrist joint (right illustration below), in order to make space for the flexion hinge, one cable joins beam to beam and the other connects the two trailing-links.
Wrist joint hinges
The Falcon's wrist joints incorporate hinges that allow the joints to flex downward even while the wings fold and extend. The drawing below shows a top view and two posterior views of the joint. The hinge incorporates two cable-guide assemblies with small roller bearings mounted in spring steel brackets to guide the cables over the bend as the joint flexes. In this way, the wrist flexion and wing folding functions are independent and can be controlled individually and simultaneously.
Here is an exploded view of the entire wing that shows more details of the joint parts.
At the start of a strong backstroke, the wrist-joint hinge will unlock (the locking mechanism isn't shown yet) and let the outer wing section drop, relieving the humerus of the task of lifting the entire wing up as a stiff unit. This causes the rotation of the humerus and forearm to accelerate. As the humerus and forearm continue to rotate, the loose hand-section is dragged along by its proximal end while being lifted and thrown backward by the air stream. During this process, the pilot pulls on the hand grip to fold up the wing, which further accelerates its angular rotation. At the apex of the backstroke, the wing flattens out and the hinge locks up again.
This combination of wrist flexion and wing-folding will enable the wing-flip (the backstroke) to happen very quickly using little or no energy.
The wrist joint hinges remain locked at all times except at the beginning of a strong back-swing. This enables the pilot to make soft, slow back-swings without unlocking the hinges and prevents the wingtips from collapsing inadvertently.
The wings will employ damping mechanisms (to be designed) to soften the shock encountered as the wing sections snap back into the locked, flat-wing configuration at the top of the backstroke.
The Shoulder-Joint and actuator unit Assemblies
The shoulder-joint accommodates three independent functions: flapping, folding and pitch/roll-control.
The power-stroke and backstroke of the flapping cycle are driven by a single hydraulic piston for each wing. The pistons are double-acting, unbalanced actuators that are driven in both directions. They produce more force in the pushing direction (for the power-stroke) and less in the return direction (for the backstroke). The pistons are 2.5 inches in diameter operating at 2200 PSI and act through 3 inch connecting rods coupled to the shoulder-joint hinge pins. The system can produce about 2700 foot-pounds of torque at the shoulder joint. If the center of lift is located six feet out from the shoulder joint and the total weight is 260 pounds (for both the craft and its pilot), it would take 780 foot-pounds of torque to hold the wings flat while gliding. The excess torque (1920 foot-pounds) is available for the power-stroke. The actuators are attached to the main axle assembly as shown in the illustration below and coupled to the hinge-pins through forward and aft connecting rods.

On each wing, a smaller pitch-control actuator is attached between the top of the forward connecting rod and the wing-root parallelogram link (shown in red).
The power source
Pressure for the hydraulic system can be supplied in various ways. We are currently evaluating several approaches. One approach uses advanced lithium polymer battery packs supplemented by solar panels. The Falcon 8a employs a free-piston hydraulic pump driven by the catalytic ignition of 90% pure hydrogen peroxide (H2O2). In either case, control will be provided by a small, on-board computer coupled to the pilot through haptic interfaces. Such systems are presently under development or being manufactured by Berkeley Bionics in Berkeley, CA. and are employed in their Human Universal Load Carrier Exoskeleton. The operator controls the exoskeleton by simply walking (or running) in a normal fashion. No other conscious effort is needed. In the Falcon, you will simply move your arms and hands to control the system and, likewise, no other conscious effort will be needed. The power source, the pump, computer and hydraulic system plumbing parts are not yet shown in the drawing.
The Haptic Interface
A haptic interface is a sensing and feedback mechanism that acts as the interpreter between a human and a robotic device. The interface senses the human's actions and returns to the human information about the state of the device. The Falcon employs two such interfaces, one for each wing. Each interface includes a hand grip, two sensors and a small parallelogram.
Wing motions are initiated by either lifting, rotating or pulling on the handgrips.
As shown below, the hand grip and sensors are mounted on a small parallelogram attached to the larger parallelogram of the wings' humerus section. The same cable-gear devices as used in the wing joints are used in the interface joints to force the interface to mimic the folding motions of the wing. The hand section of the interface tracks both the horizontal rotation (radial deviation) of the outermost wing section and imitates its vertical flexion. Because the human wrist has a lesser range of motion in the horizontal plane, the radial deviation of the interface's hand-section is less than the rotation of its counterpart in the wing.

The only connection between you and the aircraft (other than being strapped in) is through the two hand grips. You simply grasp the hand grips and move your arms and hands in an appropriate manner. The haptic interfaces detect your motions and direct the hydraulic actuators to move the wings in a corresponding manner. All the while, you receive constant feedback about the configuration of and forces acting on the wings.
Feedback is achieved in two ways. First, the joints of the small parallelograms are closely aligned with your wrist and elbow joints. This insures that your joint rotations coincide with those of the wing joints and any forces acting on the wings will be transmitted to your arms The other means of feedback has to do with how the hand grips pivot. When a flexible wing begins to swing upward, the outer section bends downward. When you lift up on a hand grip, the fact that it is hinged distally to your hand, causes your wrist to, likewise, bend downward. The two motions coincide. In this manner, you get a nearly-free feedback function with no actual linkage between your hand and the wing's hand section.
Flapping Control: An electronic sensor (shown in gray in the drawing above) mounted on the flapping-control-axis of the hand grip detects vertical motions of your arm and activates the large actuator to move the wing in the corresponding direction.
Pitch and Roll Control: Rotating the hand grips about their pitch-control axes cause the aircraft to either pitch or roll. Sensors aligned along the pitch-control axes of the interfaces detect those rotations and direct the small actuators to rotate the wings about their long axes accordingly. Turning one grip at a time causes the craft to roll. Turning both grips in opposite directions results in a more extreme roll. Rotating them in the same direction will result in pitching the nose either up or down.
Yaw Control: Like with birds, our ornithopter has no direct yaw control that is independent of motion around the other two axes. Rotation of a hand grip on one wing to initiate a roll, for example, will naturally include "adverse yaw" as the pitched wing is dragged backward due to the added drag on that side. If the right hand grip is rotated for a downward pitch, the aircraft will roll to that side and the right wing will be dragged backward, resulting in a properly banked turn.
Wing construction
The wings are built in three sections: the humerus, the forearm and the hand. Each section is a parallelogram comprised of an expanded-foam beam forming the leading-edge and a hollow trailing-rod positioned just aft of the beam. Short links at each end complete a parallelogram, connecting the leading-edge beams to their respective trailing-rods. For added strength, all three leading-edge beams include embedded hollow rods. The wings are made made progressively thinner and more flexible approaching the tip.
As a wing folds and extends, the distance between the leading-edge beam and its trailing-rod varies. A hollowed-out slot in the beam accommodates these variations.
Wing folding is controlled by the pulley-gear devices described earlier. At each elbow joint, one pulley-gear is integral to a beam and the mating gear is integral to the trailing-rod of the adjacent section. In the wrist joint, one cable connects the two beams and the other cable connects the two trailing-rods.
Each wing section is individually covered by a flexible fabric that is stretched tightest when the wings are fully extended and slacken as the wings fold. The fabric is held in its proper airfoil shape by battens. All wing battens except for the two in the hand section are attached across parallelogram arms and, therefore, their alignment doesn't change during the wing-folding process.
Fuselage
The fuselage, illustrated below, includes a pilot cage that supports the wings and actuator unit, a saddle platform, a thick, memory-foam saddle, a keel and two bulkheads enclosing a hydrogen peroxide fuel canister. The canister (shown in yellow) containing a supply of 90% pure liquid hydrogen peroxide (H2O2) is mounted in a slanted position so that, as the fuel is consumed, the remaining fuel will stay as close to the center of gravity as possible. As mentioned previously, the power source and control system, that would be attached to the fuselage, are not shown.
Not that it would ever happen, of course, but a skid is mounted to the bottoms of the bulkheads just in case you stumble during a landing.
Head Support Gimbals
To save you from getting a tired neck, your helmet is suspended from a spring mounted, gimbals-like device that allows full freedom of head motion. Your chin rests on a foam pad built into the helmet. The mounting structure behind the head doubles as a neck protection aid in case of accidents.
The Tail
The tail structure is fixed and non-articulated. It provides pitch stability for the aircraft in its default glide mode.
Using The Legs
How much one will use one's legs while flying is an open question. Primarily, you will need them to be free for a running takeoff and for landing. However, during flight, your legs will provide significant weight-shifting to maintain a stable attitude or to enhance maneuvers---especially at low speeds. During a take-off, when flapping motions are most extreme, birds noticeably shift their legs down and backwards during the backstroke and shift them up and forward during the power-stroke. The primary reason for this is to maintain level flight. Moving the wings forward shifts the center of lift forward which would result in an upward pitch, so the bird compensates by shifting its legs (and its center of gravity) forward to match the changing center of lift. The backstroke moves the center of lift aft-ward, causing the bird to dive, so, it shifts its legs to the rear to compensate. The same conditions would prevail with our ornithopter so, while flapping the wings, you would, likewise, constantly shift your legs forward and backward to compensate for the changing center of lift.
Window Shade Stirrups
If the legs are left to simply dangle, the pressure on the groin area in contact with the saddle would quickly become uncomfortable (the "dangling legs problem"). To alleviate the condition, the Falcon includes "window shade" stirrups for you to stand on to lift your body weight off the saddle---like standing on the pedals of a bicycle.
The stirrups (shown in the illustration above ) are attached by cables to spring-loaded drums. As with window shades, the drums allow the stirrup cables to extend but automatically lock to prevent the cables from winding up. If you give the stirrups a quick tug, the drums unlock and the springs try to wind up the cables. If you do this and relax your legs, the springs will draw them up into the folded position shown below. This folded-leg position is the default glide-mode position in which the craft is balanced for a stable glide.
You can extend your legs at any time. The more you straighten them, the tighter the springs are wound. The stirrup cables become fully taught just before your legs are completely straight. Fully extending your legs, then, will lift you an inch or so off the saddle and free up your legs for a running takeoff or a landing.
Taking off and landing: It's probably not possible with this ornithopter to take off in still air on level ground. Most likely, takeoffs will require a gentle headwind or a sloping surface. Simulator studies indicate that, for an aircraft with this wing area, about 200 lbs of lift is obtained at slightly over 13 MPH. Since humans can easily run at well over this speed, a running takeoff down a slope is well within the design envelope. Landing would be a matter of gliding down at the slowest possible airspeed and performing a stall maneuver at the last second, just as your feet touch the ground.