RIT will permanently shut down the RIT Wiki (Confluence wiki) on February 2, 2024. Review the Wiki Shut Down Overview for information on alternative tools, exporting content from the wiki, and other FAQs.
Team Vision for Preliminary Detailed Design Phase
The primary goal for the preliminary detailed design was to narrow down the options developed in the systems level design phase and develop a preliminary design concept (high risk) for the automated search drone.
Preliminary detailed design phase goals:
- R&D, further search for existing designs
- Develop preliminary designs to serve as a base for the detailed design phase.
- Develop test plans for the build and test phase.
- Develop basic prototypes to serve as a proof of concept
- Develop software and systems architecture flow charts
- Finalize sensor cocktail
Due to some unforeseen circumstances (COVID-19 related shutdowns), we were able to do only minimal prototyping. The following are the phase goals that were accomplished.
Goals Accomplished:
- R&D, further search for existing designs
- Develop preliminary designs to serve as a base for the detailed design phase.
- Develop test plans for the build and test phase.
- Develop software and systems architecture flow charts
- Finalize sensor cocktail
Prototyping, Engineering Analysis, Simulation
Prototyping efforts were stinted by COVID-19 and the university closure effective March 16th. Documented here is the work completed and an outline of what work needs to be done.
Propeller Selection
Choosing a motor/propeller/ESC/battery combination is an essential part of efficient system design from an electrical standpoint. In order to do this, propellers of various parameters need to be compared across a range of rotation speeds to determine a prop which can meet the thrust needs of the aircraft. After classification of propellers, motors can be chosen alongside a propeller to pair them in the most efficient way. In order to do this without purchasing several motors and ESCs, spare motors and ESCs were used which were not necessarily efficient, and only propellers will need to be purchased. A test harness was not able to be constructed due to the prototyping delay incurred by university closure.
The electrical testing system for this apparatus was simple. A small microcontroller was used to drive the ESC and motor in the propeller test apparatus and a force gauge was placed under the motor assembly, whose resistance as a function of force was sampled using an ADC. The microcontroller is commanded via a serial interface.
Airfoil Design
In order to choose an airfoil that would meet engineering requirements, we prioritized:
- 1. Stability
- 2. High lift-to-drag ratio
- 3. Easy to manufacture
- 4. Forgiving stall characteristics
- The design change to a tail-sitter configuration necessitated an effective airfoil. Since this drone would be flying at relatively slow cruising speeds, it needed to have a fairly low stall speed decently large stall angle. The general consensus was that a deep-cambered thick airfoil would be most effective at low speeds. Airfoils with flat bottoms, such as the classic Clark Y, are easiest to manufacture. Since our new design would involve two forward mounting props constantly bleeding air over the wings, stall wasn't as big of a worry as there would still be some constant lift beyond the critical (stall) angle ["Agile Turnaround Using Post-Stall Maneuvers for Tail-Sitter VTOL UAVs", Matsumoto, et al]. Six airfoil shapes were considered - the six being: Clark Y, GOE 398, Miley M06-13-128, NACA 4415, NACA 4418, and S7055. Examinations into their low speed lift coefficients, drag coefficients, and critical angles ["Airfoils at Low Speeds", Selig, et al] narrowed the field to the Miley and the S7055.
Miley M06-13-128 airfoil (http://airfoiltools.com/airfoil/details?airfoil=miley-il)
S7055 airfoil (http://airfoiltools.com/airfoil/details?airfoil=s7055-il)
They were then closely compared on their respective C_L/C_D and C_L/alpha curves. Since lift is directly related to lift coefficient (likewise with drag), this is an acceptable way to determine lift effectiveness.
Lift coefficient vs. angle of attack for the Miley M06-13-128 (http://airfoiltools.com/airfoil/details?airfoil=miley-il)
Lift coefficient vs. angle of attack for the S7055 (http://airfoiltools.com/airfoil/details?airfoil=s7055-il)
Ultimately, the Miley was deemed to be too volatile and its shape was not as easily manufactured as the S7055. The S7055 exhibit greater consistency in lift capacity, and had better lifting capabilities at low Reynold's numbers (i.e. low speeds - the light blue line corresponds to a Reynold's number of 50,000: the lowest readily available analysis number. Later analyses showed our drone to operate in Reynold's numbers four times that, even when flying at speeds lower than desired.). The S7055 was chosen as the main airfoil to move forward with.
Along with the main airfoil choice - the S7055 - a symmetric airfoil was considered. A symmetric airfoil, when at 0 degree angle of attack, produces zero lift, as the air flows over the top and bottom surfaces at the same rate. For this reason, it is used in some research tail-sitters to facilitate hovering vertically ["Full Attitude Control of a VTOL Tailsitter UAV", Verling, et al]. The S7055 was analyzed for lift effects when in vertical hover mode for comparison.
Ultimately, it was shown that the S7055 would still be able to hover with minimal elevon trimming. The S7055 also produces positive, non-zero lift at a 0 degree angle of attack. These qualities make it a better airfoil choice than the symmetric E168.
In order to calculate lift, a wing area and chord length must be known. These parameters were outlined prior to any lift calculations as such:
Please note: the planform area was calculated as if the wing was completely flat. This is obviously somewhat in error, as the wing is inherently curved, but assumed a completely flat wing allows for these calculations to be worst case scenario. Likewise, the elliptical curves of the trailing edge and wing tip were considered to be straight hypotenuses. This means the area we are doing calculations with is smaller than the true area, but it allows for a worse case scenario analysis. Induced drag - as well as lift - does increase with area, but its effect is negligible, especially compared to speed. For an aircraft the size of ours (single wing area is <1 m^2), increasing speed will increase drag and lift by orders of magnitude more than increasing area could.
The wing size was based on existing surveying drone proportions and combined with an elliptical wing accoutrements. Elliptical wings are well known for having an elliptical and stable planform lift distribution.
Above is a comparison of an elliptical planform wing's lift capacity as a function of its span and that of a rectangular planform wing. (https://www.sciencedirect.com/topics/engineering/lift-distribution)
Elliptical wings are harder to manufacture, so our planform has a straight leading edge - this also makes it easier to mount rotors forward on the wing. Each wing has a span of 32.5 " or 0.8255m. Long, wide wings will provide greater lift capacity and stability.
A lift analysis was executed on the wing planform, at a cruising speed of 35 km/h or 21.75 mph. This is 5km/h lower than the targeted 40km/h cruise speed engineering requirement - this was done to ensure sufficient lift at a stall speed lower than expected. The analysis was run for air at 1 atm and 0 degree C, 15 degrees C, and 30 degrees C for a single wing geometry.
The graphs show that lift performance increases as temperatures drop, insinuating that our UAV will be able to operate in more inclement weather. Additionally, even for a speed lower than necessary, the lift force of the wing is far above the induced drag force of the wing. Most importantly, however, is that this airfoil-planform design allows us positive lift force even when the angle of attack dips below zero, allowing for increased maneuverability.
Emergency Release System
- Upon information from the on board computer the parachute will be deployed in the event of failure
- Servo will rotate head and move arm off top of parachute canister
- Spring loaded Parachute will be deployed
- This system will be reusable
- 36" canape can hold 4 lbs (48" would be ideal size to hold roughly 4.5 - 4.7 pounds)
- Slow decent speed to mitigate damage
Feasibility: Prototyping, Analysis, Simulation
Emergency Release
- Most existing designs on the market were very similar Examples are shown below:
|
|
|
|
From these existing designs, the key features extracted were:
- Slim Canister
- Attached Servo to hold down lid
- Compression spring inside
- Rolled up parachute
Overall, the parachute design was pretty straight forward just needed a vessel to hold the parachute tightly wrapped up with a spring underneath and a servo holding the lid down. (Preferably a metal servo strong enough for this task).
Once the servo moves its head the spring will decompress deploying the parachute for our needs. These simple components lead to our initial design of the Emergency Launch
Camera Gimbal
Significant design modifications have mitigated the need for a gimbal given what is currently known about the design. Some form of a gimbal or vibration dampener may be developed if deemed necessary.
Sensor
Distance Sensor
Measures distance using lasers and time of flight measurements. Cheap and accurate solution to calculating distance. We plan on including three of these devices:
- Belly Sensor: Detect landing transition
- Two Tail Sensors: Detect landing surface (check if its flat)
Camera
Three cameras are being considered for use:
- Front Camera: FPV for Pilot
- Belly Camera: Surveying while gliding
- Rear Camera: Surveying while in VTOL
The piloting camera can be satisfied with a cheaper FPV camera that provides the pilot with just enough resolution to drive. However, the surveying cameras will need to be of higher resolution. Additionally, a more powerful board will be required to drive these 1-8 MP cameras. Thus, we are considering using a RaspberryPi Zero and the 8MP RaspberryPi camera.
Positional Sensors
Barometer - Measures atmospheric pressure.
Gyroscope - Measures orientation and angular velocity.
Accelerometer - Measures acceleration.
Magnetometer - Measures magnetic forces.
These sensors can come as a package on a breakout board, providing a cheap efficient solution for all these sensors.
GPS
Measures the position of the drone using satellite positioning. Needed to determine the location of the drone and the location of any person located during flight.
Airspeed Sensor
Measures airspeed using a pitot tube. This device will allow for stall detection and airspeed monitoring. However, it is extremely expensive costing an average of $50 per device. If we know the aircraft can not stall, we can avoid having this sensor.
Rangefinder
Measures the distance from the observer to a target. This sensor will not be included in the package because of bulk and expense. Its capabilities can be replaced by the distance sensor for a much lower cost.
Controller Choices
- K64 from NXP is head-to-head with Raspberry Pi for the central controller.
- Smaller "slave" controllers are being tenatively selected, though the number and complexity required is unknown. AVR microcontrollers are familiar and readily available, and may be used for these purposes if they can be shown to be sufficient.
- Small ARM cores for slave microcontrollers are also under consideration.
- UARTs will serve as the main network backbone for inter-controller communications (subject to change)
- FPGAs are under consideration as a "last resort" if something cannot be done quickly on a microcontroller or extreme precision is required for some operation.
RaspberryPi Zero
| Pros | Cons |
|---|---|
| Allows for cheap camera ($20-30 vs $60 (CMOS Camera + Respective Board) vs $100+ (Point and Shoot) | Power hungry in comparison to microcontrollers |
| Ease of programming (Higher Level Languages), less manual optimization | Not real time |
| Potential for image analytics | Linux operating system overhead and complexity |
| Has a filesystem for storing error messages and other logs | 1 UART |
| Highly debuggable, potential for wireless connection | Filesystem corruption possible |
K64
| Pros | Cons |
|---|---|
| Familiarity | Cameras will require separate power hungry processor and will be more expensive |
| Multiple UARTs | Low overall data throughput |
| Simpler | Limited multitasking / computational ability |
| Extremely powerful timer/PWM generation subsystems | Relatively difficult to debug |
| Powerful interrupt priority scheme | Proto-board is heavy compared to RPi Zero |
| Zero overhead (Bare metal) | Small RAM size |
Telemetry
Telemetry components are being selected later, as they are low-risk at the current time. It is unlikely that a solution will be developed, one will instead be purchased and adapted to the project needs.
Updated Engineering Requirements
Drawings, Schematics, Flow Charts, Simulations
High-Level System Architecture
Highlights:
The design is logically modular, but the final realization of the logical design may involve different hardware and software separations than the above.
Each unit is individually testable in simulation.
Physical separation is under consideration, as are the processing systems for each logical component.
High-Level System Data Flow (Control Systems)
Highlights:
Logical components can be further broken down into algorithms and processes.
Data resulting from these processes are presented over an interface which is not specified due to uncertainty in the physical implementation
"Interfaces" are easily replaced and tested when the underlying process is not being debugged at the same time, thus making this design robust to architectural changes.
Power Distribution Scheme
Modular Power Supply Scheme
Highlights
Single supply that can output both 5 V and 12 V.
Less noise is generated by this design. Also fewer components.
All components in grey will be protected using capacitors to prevent damage from electrostatic discharge.
Two Power Supply Scheme
Highlights
Two separate power supplies that output 12 V and 5 V.
More noise due to the presence of two power supplies.
All components in grey will be protected using capacitors to prevent damage from electrostatic discharge.
Single Supply w/ Power Distribution Board Scheme
Highlights
The main power supply (20 V in schematic but could be just 12 V) connected to a power distribution board.
Less noisy when compared to two separate power supplies, but more components.
All components in grey will be protected using capacitors to prevent damage from electrostatic discharge.
Course Correction Software Flow
Highlights
Two types of telemetry data could be received, both serviced by the central control module
Central control is in charge of routing and servicing specific modules
The grayed out area may be all under the same processor or separated into different processors, depending on microcontroller choice
Debug and Status System
Highlights
All subsystems are queried for their status by the central processor, these statuses are written into the debug logs
Depending on the type of central processor, the probe may be a debug probe or simply a device to retrieve a file
Status codes are binary encoded to save space and debug logs can be parsed by a program to match these codes to a specific status
Motor Control Flowchart
Airframe Design Choices
VTOL Tail-sitter test bed, based on Andy Meyer's current personal drone.
Custom tail-Sitter design with landing struts incorporated into fuselage.
Custom tail-sitter design with landing struts as winglets. These also serve as housing for rangefinders.
Emergency Release Design
Using the existing designs as reference we developed a parachute emergency release to look something like the images above. The emergency release should be able to fabricated by 3D printed and then adding the external components (Servo, Parachute, Spring) after. Overall, a simple effective design. Since the air frame is not finalized attaching the chute is not yet included in the design.
Part Drawings: (Note some dimensions will change)
Bill of Materials (BOM)
Note: The BOM plan covers the cost of a singular drone without repair and replacement cost. That budget will come out of the total budget allocated for this project of $1500. This is not shown due to the fact that the goal for one drone is $600. After purchases start being made, total budget projections can be accounted for.
Test Plans
Test plans are shown here as a list. They may be accessed under "Test Plan Home" on the P20123 space as well. This list is automatically generated and will update as test plans are created.
- S01 Test Plan
- S03 Test Plan
- S04 Test Plan
- S05 Test Plan
- S06 Test Plan
- S07 Test Plan
- S10 Test Plan
- S11 Test Plan
- S12 Test Plan
- S13 Test Plan
- S14 Test Plan
- S15 Test Plan
- S19, S20, S21 Test Plan
- S22 Test Plan
- S23 Test Plan
- S24, S25 Test Plan
- S26 Test Plan
- S27 Test Plan
- S29 Test Plan
- S31 Test Plan
- S36 Test Plan
- S37 Test Plan
- S38 Test Plan
- S39 Test Plan
- S41 Test Plan
Risk Assessment
Risk | Description | Risk Type | Likelihood (0-5) | Severity (0-5) | Importance (L x S) | Action(s) Taken |
|---|---|---|---|---|---|---|
COVID-19 | We are currently facing a pandemic that has everyone social distancing. | Safety | 5 | 5 | 25 | Zoom meetings instead of class meetings. Team members must rely on online communications to get work done. |
Overbudget | Budget is limited - no funding outside of team members and RIT | Resource | 3 | 5 | 15 | Avoid |
Fabrication | Machine shop/Construct tools could be busy, or available building resources could be limited | Time | 3 | 3 | 9 | As soon as a preliminary design is settled upon, start to source materials and fabricate parts as they become finalized, not once all parts are finalized |
Software (telemetry) | Telemetry is difficult, and difficult to debug. Significant technical challenge. | Technical | 3 | 3 | 9 | Start early, plan ahead for this challenge, since it will be necessary during testing. |
Injury | Someone could be hit by the drone, or cut by the spinning prop | Safety | 2 | 5 | 10 | Have a designated drone spotter when testing, mark dangerous area on drone with labels, have a testing plan with steps clearly outlined to ensure personnel safety, only test drone in areas away from residential areas |
Drone Protection | Effective crash mitigation tech may be expensive or technically difficult to implement, affect CR's and putting drone safety at risk | Resource | 4 | 3 | 12 | Choose a concept which has crash protection built in. Ensure pilots are trained before flying the real aircraft. |
Efficient Wiring | Small problems in wiring and electrical design can become troublesome quickly. | Technical | 3 | 3 | 9 | Consider current flows, do initial testing. Overbuild subsystems likely to cause problems (engine power subsystem, for example). |
Testing | A quick, accurate, and safe way to test drones is critical given that we cannot test on campus | Technical/Time | 5 | 4 | 20 | Choose a concept which we are capable of testing. |
Flight Time | The drone should be able to maintain flight long enough to test fully and not worry about sudden loss of power mid-flight | Technical | 2 | 4 | 8 | Choose a concept for which we can optimize power consumption and therefore flight time. |
Limited Number of Drones | Only so many testing drones can be constructed with budget constraints | Resource | 3 | 3 | 9 | We will only be able to build one drone, but we have two indoor aircraft that may be used as trainers. |
Loss of Team Member | Reduced work capacity due to missing/absent team members. | Resource | 5 | 1 | 5 | Mitigation plan is to reduce our dependency on custom mechanical parts. (Ben is leaving) |
Customer Redirection | A new customer is added, thus changing CRs | Time | 1 | 4 | 4 | Mitigated. At this point it is not feasible for the team to take on a new customer. |
Customer Absence | No official outside customer is identified, leaving us with speculated CRs and little expert knowledge | Information | 5 | 2 | 10 | We have been continuing our design validation and analysis techniques. |
Extended Winter | Inclement conditions make flight testing difficult | Time | 2 | 4 | 8 | Find resources where we can reserve indoor flying space, even if small. |
Unexpected Roadblock | Unforeseen technical hurdle, miscommunication or serious bug not found in testing | Technical | 5 | 3 | 15 | Plan project. Break large tasks into smaller ones until they can not be broken down further. Ensure communications are clear and quick. |
Team Communication | Wealth of information spread across platforms and people could create discrepancies in planning | Personal | 5 | 1 | 5 | Communicate in advance of personnel absences, after every meeting post in group chat what assignments were and the expected due dates so there is accountability |
Phase 4 Preview
Team Goals
As a team, we would like to make final decisions on all aspects of the preliminary detailed design in order to be able to develop a single design that meets the maximum number of CRs and ERs possible. We would also like to finalize the bill of materials (BOM) required for the build and test phase. Due to new purchasing restrictions and other general constraints brought on by the COVID-19 pandemic, prototyping is going to be difficult, however, we are going to do our best to develop proof of concepts for our designs.
Individual Goals
Atulya
Select the best power scheme to use.
Develop detailed schematics for the chosen power scheme.
Choose specific components for the chosen power scheme.
Develop more elaborate test plans for the various engineering requirements.
Look into facilities where we can test.
Piers
Help finalize motor and propeller selection
Help finalize servo selection
Finalize sensor cocktail component selection
Camera selection
GPS selection
Accelerometer/Gyroscope/Barometer/Magnetometer package selection
Help select central control processor
Help finalize debug system architecture
Sabrina
Finalize cost of all expected materials
Work with Andy on specific test plans, specifically for software testing
Explores software prototypes with selected sensors
Aid in finalizing sensor cocktail component selection
Create purchasing plan (when and where to purchase materials from)
Mutahir
Help Finalize Airfoil/Craft Design
Finalize emergency release system
Help Finalize/Design Electronics holder to keep everything together and safe
Help Finalize material for airfoil and frame
Ensure all electronics fit with frame.
Ben
Work with Mutahir to finalize design.
Help develop CAD model to be used in simulations.
Run more analyses for lift capacity in inclement conditions.
Develop an plans to construct wings, etc. out of foam core to build a simple proof-of-concept prototype.
Andy
- Work on MCU selection and main controller choices
- Design the physical system architecture
- Concrete-ify logical architecture with a software and hardware architecture
- Create Git repositories for code, set up BAT-automation and development environments for all controllers chosen
- Develop debugging plans and protocols further, implement some basic testing code.
- Work with mechanical on emergency release and electrical interconnects for submodules and releasable black box.