Closed Quarter Aerial Robotics (CQAR)
 Figure 1: CQAR prototype |
 Figure 2: Surveillance image take from CQAR prototype |
 Figure 3: Recreation of image taken with digital cam |
Figure 1: CQAR prototype |
Figure 2: Surveillance image take from CQAR prototype |
Figure 3: Recreation of image taken with digital cam |
Indoor aerial robotics is a new and emerging field of study and is the central focus of my research. I am currently working on a project entitled Closed Quarter Aerial Robotics (CQAR) that has applications such as surveilling/monitoring large indoor areas to safe keep stadiums, warehouses, subway tunnels and train stations. The CQAR prototype (designed by Gordon Johnson) pictured above weighs 27 g (app. the weight of 3 quarters) and flies at about 5 mph. Below are links to a parts list and step-by-step instructions on how to build your own slow flyer:
Milestones in aerial robotics have been achieved in recent years but mainly flying outdoors or limited to lab workbenches. By contrast, aerial robots flying in closed quarters are rarely discussed in the literature. There is a need to surveil or monitor large indoor areas to safe keep stadiums, warehouses, subway tunnels and train stations but is human-intensive and time-consuming. Col. John Blitch, director of the September 11th robotic search-and-rescue effort commented at a plenary session that ground-based robots can be used indoors but have “limited mobility and field-of-view”. Rotary-wing aircraft, like helicopters studies by Sukhatme and Sastry and four-prop vehicles by Mahony rely on global-positioning systems and hence cannot be navigated indoors. Blimps studied by Ostrowski can be employed but often are too wide to fly through doors. Furthermore, the focus has often been in overcoming difficulties in constructing an airworthy vehicle than in examining and implementing autonomy. The USC team led by Sukhatme is now on their third generation of robotic helicopters. Recently, they claim to be the first team to achieve vision-based autonomous landing. Additionally, most unmanned aircraft in the military are remotely piloted with development focus on human-computer interaction and teleoperation rather than autonomy. DARPA’s micro air vehicle (MAV) program prescribes flight performance (20 to 40 MPH) demands that are unsuitable for closed quarters. Today's electric motors, Lithium-Poly batteries, fast-embedded microprocessors, miniature radio-controllers and carbon-fiber materials permit designing small aircraft that is less than 10 grams and has a 30 cm wingspan and flies less than 2 m/s in closed quarters (see Figure 1). I have had experience in applying such components in designing a kite that carries a teleoperated camera. I believe the components can be applied to construct an autonomous indoor flying robot, retrofitted with sensors and embedded with algorithms, like collision-avoidance, that are well-known in the ground-based mobile robot literature
The gaps in the existing knowledge are: one, absence of low-speed aerodynamic airfoil data which would permit indoor flying; and two, sensor-based algorithms have been applied to ground robots that have two degrees-of-freedom, they have not been modified nor assessed for aircraft which possess six degrees-of-freedom. The Reynolds numbers is the critical measure of whether something will fly. Wind and water tunnel tests are typically done for fast flying aircraft. Also documented are tests on insects and micro air vehicles for Reynolds number ranges of 1,000-to-10,000 and 10,000-to-100,000 respectively (Mueller [11]). Absent is data required for indoor flight, namely Reynold’s numbers around 20,000. Closed quarter aerial Robotics (CQAR) requires flying at very low speeds (less than 2 m/s) and without this Reynold’s number data, designing airfoils will remain ad hoc and analytically impossible to optimize.
The CQAR prototype is ideal for indoor flight because of its low flying velocity. It's minimum velocity (with Cl assumed to be 1.0) was calcuated at 2.35 m/s (5 mph) using the formula below:
I am currently constructing a CQAR prototype and plan to use it to carry out a lot of experiments in my research. The fuselage and wings are constructed with carbon fiber rods and the wings are covered with mylar. It is a 27 gram slow flyer with a wing span of appx. 19 inches and a wing area of 123 sq inches. The weight distribution, along with some components, of the CQAR prototype are shown below:
 Figure 2: Component weight distribution |
 Li-poly battery.jpg  Toytronix motor.jpg |
 Propeller.jpg  Receiver.jpg |
I knew I needed an aircraft that weighed less than a 100 grams, capable of flying at speeds less than 8 mph, and with a wingspan of less than 1 meter. So I created a table with several weights and wing areas and calculated the corresponding velocities. Then, I calculated the room length using a minimum of 5 seconds for the control system to react. The room length was plotted against the weight. The net result is that by knowing the size of the room I wanted to be able to fly in, I could then look at the graph and get good idea of the weight and wing area required for the aircraft to achieve this.
From the graph, it can be seen that the 27g CQAR prototype can be flown in a minimum room size of 144 square meters (12 x 12). This is equivalent to 1/3 of an NBA court. Furthermore, if it were physically possible for the CQAR prototype to carry a 12.5 gram payload while keeping the wing span and wing area the same, the minimum room size would increase to about 14 x 14 meters. However, if it were necessary to keep the minimum room size constant, then the wing span, thus the wing area, would have to be increased (see light blue graph). The wing of the CQAR prototype is an oval, so when calculating the wing area (s), it can be thought of as a rectangle and 2 semi-circles.
Design Guidelines
These guidelines apply mostly to models with wingspans in the range of 24 inches or less and for models with overall weights of 4 ounces or less, but the principles are similar for larger/heavier models. In general, the goal is to swing the largest diameter prop at the lowest current drain, with the lightest battery pack, for the longest flight duration.
1. Start with a very small and light receiver.
2. Add an ESC to provide the voltage for the receiver and servos or actuators.
3. Choose actuators or servos, depending on weight and price factors.
4. Choose a motor for the model using reduction drives from 3, 4, 5, or 6:1.
Motor Selection: Electric motors typically do not like to run slowly. When they do, although they provide enough power to fly a model, they require a lot of current and, thus, larger batteries. But if we can let an electric motor run at high rpm, where its efficiency is best, it can provide a lot of torque if we transmit it through gearing that lets the motor run fast while the prop runs at a much lower rpm. For example, a small Mabuchi N-20 motor does best on about a 2-inch-diameter prop. However, when fitted with a 5:1 gear drive, it will swing a 4-inch or larger prop and fly a larger and/or heavier model. The amount of thrust required, which can be determined from the thrust to weight ration (T/W) from below, dictates what type of motor to select. Typically, a motor thrust of about 1/3 the plane's total weight is required for horizontal flight.
5. Choose a prop, depending on the motor chosen, ranging from 3 to 7 inch diameters.
Propeller Selection: The propeller selection is based on what motor you choose. A propeller typically has 2 dimensions (160 x 120 cm). The first dimension is the propeller diameter and the second is the pitch. A propeller with a 120 cm pitch moves forward 120 cm for every 1 revolution of the blade. The motor you select usually gives a suggestion as to what size propeller should be selected for full drive. Using reduction, a larger propeller can be used which allows your vehicle to fly slower. The formula below can be used to estimate what your propeller pitch should be:6. Choose a battery and use however many cells are necessary to provide the voltage the motor and prop combination calls for.
Pitch (m/rev) = V(m/s) / [ Motor speed(rps) x Gear reduction ]
Typically, the pitch should give a theoretical propeller speed of at least twice Vstall.
Battery Selection: The battery selection is based on the current draw from the motor and propeller. This can be measured by assembling your motor, gears, and propeller and then measuring the motor current. Then, having a good idea of how long you would like your vehicle to remain airborn, the battery can be selected using the formula below:7. Choose a wing loading at which you’d like to operate.
Charge(mAh) = (estimated flying time [hrs]) x (current draw [mA]) => (.27 hrs) x (500 mA) = 135 mAh battery
For example, with a 3-ounce model and a wing loading of 4 ounces per square foot, the wing area needs to be 108 square inches; 85 square inches for a 5-ounces-per-square-foot wing loading. In general, the larger the model, the greater the wing loading can be without seriously eroding flight characteristics. The CQAR prototype has a wing loading of appx 1 oz. per square ft.
New Wing Design to Carry Payload
The goal of CQAR is to have an indoor autonomous flyer. This requires the CQAR prototype to have payload specifications large enough to carry sensors with a combined weight of 12 1/2 grams. Assuming that we would want to keep the same wing loading value, then the equations would come out to be:
A standard doorway is 3 feet wide. So to keep the wing span at a minimum, hence maneuverability at a maximum, the chord length would have to be increased in order to get the wing area up to 189 square inches. Rearranging the wing area formula from above, the wing span is equal to:
Increasing the wing chord from 7 to 9 inches will give us a new wing span of 23 inches. Therefore, when flying through doorways, CQAR will have a 6 1/2 inch leeway on both sides. Furthermore, because the lift force has to balance the weight during cruise flight (L=W), the required lift coefficient can be caculated using the formula:
CG Determination
The non-negligible weights, along with the locations of their respective individual locations measured relative to the nose of the airplane are shown in Figure 5. The effective center of gravity is calculated by summing the moments about the nose and dividing by the sum of the weights. The result is:
 Figure 5: Sketch for moment calculation about nose |
 |
Power Required
The required power for the CQAR prototype to takeoff can be calculated and will give a better idea of what battery to select. The values below are going to be used in the power calculations:
Gross weight (W) = 23.4 g = .0516 lb Wing area (s) = 123 sq. in. = .8542 sq. ft. Density (sea level) = .002377 slug/cu. ft. CLmax = 0.90 (appx.) Takeoff distance (sg) = 3 ft (appx.)
I assumed a takeoff distance of about 3 feet and ignoring the clearance of obstacles after takeoff. For this case the thrust to weight ratio can be calculated using the equation below:
Now the required takeoff power can be be calculated:
