Team Vision for Detailed Design PhasePLANNED
- Close action items from Preliminary Detailed Design
- Validate Customer and Engineering Requirements
- Conduct appropriate level of analysis to verify functionality of design with input from subject matter experts
- Finalize Bill of Materials with suppliers, costs, and possible areas of improvement
- Prepare budget proposal
- Detail Software and finish development
- Assess remaining risks and develop mitigation plans and triggers for MSD II
- Create hardware/subsystem level test plans
- Create Engineering Requirement validation test plans
- Update drawing package to incorporate specs and revision control
- Develop project plan for MSD II
AREAS OF CONCERN
- Software Development - Implementation into computer and hardware interfaces
- Directionality of transmitters
- Effect of budget on testing and ability to reach requirements
Problem Statement and Project Summary
- The Amelia Project’s goal is to develop a drone system unlike any other on the market. The drone will be equipped with the ability to survey the entire 360 degree sphere around the drone through a camera mounted to the underneath of the drone. This video will be fed live to the drone pilot wearing a virtual reality headset at a base station through a WiFi transmitter-receiver pair connecting the drone and the base station computer. This will allow the pilot to see from the drones perspective and give a better sense of where possible hazards in the environment for the drone may be. The expected result is a functional prototype that is presented to Lockheed Martin to demonstrate to the management the effectiveness of these technologies when combined together.
- For this goal to be achieved there are some important
- Low latency between the video and headset to make controls feel smooth and responsive
- High signal data throughput to send large images that will look clear on the headset
- Noise reduction to have a clean and artifact free video
- Following FCC and FAA regulation
- Since the initial design plans, some important sections were redesigned and improved. Opting for a 360 degree full hemisphere rather than the initial 360 degree lower hemisphere adds visuals to what would have been an empty space but at the cost of doubling the required throughput. Another important redesign included upgrading from small raspberry pi-like small boards to handle transmission and receiving to company made transmitter and receivers that are designed for high throughput at high distances. The importance of high quality transmitters was learned from meeting with subject matter experts (SME). All design choices that get changed cause a lot of reworking that ripple through the project and show the importance getting every piece of information you can when making choices.
Hardware / Subsystem Test
To view the whole test document: Hardware Test Plans
Heat Transfer AnalysisPer the instructions of Dr. Stevens, the camera is the limiting factor and modeled as a fin where the mounting area is treated as being adiabatic. Below are the conditions analyzed:
- Maximum operating condition: q = 8W
- Maximum usage temperature: T = 40C
- Maximum storage temperature: T = 60C
- Steady state analysis
- The maximum ambient temperature with the current configuration is 87F
- To increase the temperature, the mounting area may be reduced, or a heat sink may be installed which has a mass of approximately 100g
For the full analysis: Heat Transfer Analysis
We want to transmit 4K video live from the Ricoh Theta V to the base station. According to the Ricoh Theta V specifications listed on the Ricoh website, the Ricoh Theta V livestreams H.264 encoded 39 FPS 4K video at 120 Mbps (megabits per second) using the USB cable. We intend to use this interface method for our application.
Between the radio transmitter we intend to use and the Ricoh Theta V sits a USB OTG to ethernet adapter. This is required to livestream video using the Ricoh Theta V’s Android based “plugin” system, since no developers seem to have been able to get access to the USB host controls on the Ricoh Theta V. This adapter makes the USB connection appear to be an ethernet connection to the Android operating system on the Ricoh Theta V. USB OTG is a USB 2.0 based connection standard. While USB 2.0 had a theoretical 480 Mbps throughput, in real world scenarios (as a result of bus access constraints) USB 2.0 speeds are closer to 280 Mbps. At half of the expected USB 2.0 throughput, we should still be able to transmit our 120 Mbps video stream. The ethernet cable from the ethernet adapter is then connected directly to the radio transmitter.
We intend to use a PtP (point to point) WiFi / radio connection to stream our encoded video from the drone to the base station. We intend to use the Ubiquiti Nanostation 5AC as the transmitter mounted on the drone. According to the NanoStation 5AC specifications listed in the User Guide the NanoStation 5AC has a realistic throughput of 450 Mbps, and a range of 15 km when using the highest frequencies and 256 QAM modulation. The NanoStation was selected as the transmitter since it has a relatively large receiving cone, suitable for the vibrations on a UAV platform. If the built in antenna underperforms, we could always purchase a high gain, omni-directional antenna for the transmitter.
On the ground station we intend to complete the PtP link using a NanoBeam 5AC Gen2. According to the specifications in the User Guide, the NanoBeam 5AC Gen2 has a range of 15 km and a throughput of 450 Mbps. Compared to the NanoStation 5AC, it has a narrower FOV. This is fine for a ground-based receiver since we can accurately aim the NanoBeam and don’t need to worry about vibrations.
Even at half of Ubiquiti’s “real world” throughput of 450 Mbps, there should be no throughput issues transmitting the Ricoh Theta V’s 120 Mbps 4K video feed. The 120 Mbps video is fed through the USB to ethernet adapter to the NanoStation, which can reportedly beam data 15 km at 450 Mpbs. Then at the base station the NanoBeam communicates with the NanoStation, and has similar specifications. Our engineering requirements only require us to transmit our video 0.25 miles, or 0.4 km, which should allow our throughput to approach the advertised NanoStation / NanoBeam throughput rates. The caveats are there may be issues regarding range and signal quality given the test environment and the alignment of the NanoBeam and NanoStation. Fortunately these issues can be alleviated by purchasing new antennas and by consulting SMEs.
The above gain diagrams are provided by Ubiquiti and characterize the gain of the NanoStation 5AC that will be mounted on the drone. As shown by the graphs, signal reception directly behind the transmitter is poor or nonexistent. We will perform testing and analysis to make sure the transmission hardware is properly oriented for maximum signal throughput. In the event that we receive a budget increase and the testing yields inadequate signal reception with the NanoStation 5AC, we could opt to use an omnidirectional antenna. The gain diagrams for the omnidirectional antenna we would use can be seen below.
Software Design and Benchmarking
- CPU: I7-4790 vs. I5-8300H
- 0% difference (userbenchmark.com)
- Software Languages and Programs-
- C# in Unity design
- Batch file language for video storage and piping
- Internet Protocol for WiFi signals
To track the payload we are adding to our drone as well as the overall weight, a chart containing the mass of every component on the drone was developed. This allows us to see a conservative estimate of the total weight of our drone, and ensure that it does not exceed the maximum payload specification of our chosen drone.
Mounting Bolt Calculations
The mounting holes on the DJI S900 drone that we purchased are M1.6x35. Due to the small size of these bolts, it was relevant to perform calculations to ensure they do not fail under shear stress during the worst case flying conditions. After conducting the calculations, it was discovered that the bolts will not fail under the worst case loading by a very large margin.
Center of GravitySince a drone is a very lightweight aircraft, changes around its center of gravity can have significant effects on the way it is able to perform. We saw this as a relevant future concern since we will be adding extra payload to our drone. We decided to seek help from a subject matter expert in flight dynamics, Dr. Crassidis. To ensure the center of gravity does not change very much, the payload should create as little of a moment around the drone's center axis as possible. To account for this, the harness was designed with modular mounting slots.
Subject Matter ExpertsWe encountered various topics and issues during the design process that we did not have a lot of knowledge about. For these design hurdles, we sought out subject matter experts in various fields to assist with our design and overall understanding of the hardware we would need and the system we were dealing with. The professors that we met with were Dr. Indovina who is an SME on wireless transmission, Dr. Steven's who is an SME on heat transfer, and with Dr. Crassidis, who is an SME in aviation dynamics and controls.
To thoroughly document all of the hardware that we will be using for this project, drawings for each piece of equipment have been made. These drawings document important specifications for each item, give a visual representation of the parts that will be used, and detail the necessary specifications needed for each piece of hardware.
- Low price, high power laptop that is an "Oculus Ready" computer
Bill of Material (BOM) & Budget
With the current budget we can achieve the original engineering requirements, but with some limitation that limit the scope and applicability of the project. To be clear, with the current budget we can achieve the primary engineering goals: 0.25 mile range, 20 minutes flight time, full 360 degree video stream, interactive VR experience.
However, there are limitations to the current system, as well as budget related impediments that could have a large impact on delivering a fully functional and reliable system by the end of next semester.
- Limited transmission FOV: The Ubiquiti NanoStation 5AC has a narrow FOV, intended to operate as a part of a point-to-point receiver/transmitter mounted as part of a communications relay atop a tall pole. Our use case is slightly different, and would benefit from a omnidirectional antenna. The NanoStation 5AC was a decent compromise, so long as we keep the NanoStation 5AC pointed towards our base station. This can be accomplished a number of ways (actuator on the drone or modifying the flight controller to keep the transmitter aimed at the ground station), but the performance will likely be spotty. For demonstration purposes, the system would likely be flown in a straight line, with the NanoStation 5AC pointed directly at the ground station.
- Long down-time between flights: We have a number of power systems that we need to have charged in order for flight. We have a power bank for the ground station, a power bank for the transmission system on the drone, and a battery for powering the drone itself. The current budget barely has space for a single one of each of these batteries. The current drone battery has the power for a 20 minute flight with a full payload. The power banks each would take roughly 5 hours to charge, and the drone battery will take roughly 80 minutes to charge. This is a huge hindrance during flight tests, and basically means we will only be able to perform one test flight a day, greatly slowing down testing and validation.
- Small Spare Budget: In the event of issues, we have less than $200 to handle the problem, and test hardware hasn’t even been built in to the budget yet. In the event of a piece of hardware failing to perform up to spec or breaking, we will be high and dry, and likely be unable to complete the project. We have basically no room for failures or system redesigns given the current budget. And we don’t have the funds for test equipment or a noteworthy Imagine RIT demo.
All of these issues can be remedied with a budget increase. Our current budget does not give us the leeway for failures or redesigns, and requires us compromising on certain objectives. It also greatly slows down our development process, meaning that the resulting system will be less polished. A budget increase would allow us to solve the issues in the following ways:
- Omnidirectional Antenna: Rather than using an Ubiquiti NanoStation 5AC with its narrow FOV, we could purchase a Rocket 5AC Lite and an AirMax Omni. This system will provide a truly omnidirectional passive transmission solution for the drone. This would mean the drone would be free to move in an open field. The system was previously outside our budget. In total it would cost us $250.
- Extra Batteries and Power Banks: In most professional drone test systems multiple fully charged drone batteries are kept on deck, so that test flights can be done in series with no down-time. Each of these drone batteries cost $220. This allows a gamut of tests to be run and prioritizes the time of engineers. We would also want to purchase 1 or 2 extra power banks since the charge time on those is rather long, although we won’t be draining them very fast.
- Spare Budget: Included in the budget increase is the ability to purchase spare cables, fasteners, test equipment, and more importantly, we have money in the event of accidents. If a propeller breaks, we have money to buy a spare one. We also have money to make a nice demonstration for Imagine RIT.
While attempting to satisfy our budget constraints, a few steps were taken to minimize costs. Rather than using DJI’s N3 flight controller, we are using a more customizable and open source solution by combining a PixHawk 4 Flight Controller and a PixHawk 4 Neo-8MN GPS. This measure saved approximately $100. We are also using a relatively cheap radio controller (the FrSky Taranis Q X7). For the transmission system there are alternative companies that build backhaul networking hardware similar to Ubiquiti, but they are all comparatively priced, and we have a local SME that explicitly recommended using Ubiquiti hardware.
Unfortunately, there are certain high-ticket items that we are unable to compromise on.
- Drone: The most expensive element on our BOM is the DJI S900, which is one of the few affordable drones capable of carrying our payload. There are drones with similar payload capabilities at lower costs, but they are all made and shipped directly from China, and have small (or nonexistent) support communities. DJI has a large community that can help supplement their official support.
- Base Station Laptop: The base station laptop was also a large purchase, but the laptop we have selected is the cheapest “VR Ready” Occulus certified laptop we could find. We have a series of tests that we are currently running to see if we could run the base station application on cheaper laptops, but so far the tests are showing choppy video playback.
- Camera: The market for 360 degree cameras is still small, but is growing fast. Out of the seven different cameras tested, the Ricoh Theta V was the only camera that offered a 4K live video feed. Fortunately, the Ricoh Theta V is competitively priced with the other cameras we reviewed, and most importantly, has an active developer community we will be able to work with us. There is even demo code on the developer site showcasing the Ricoh Theta V livestreaming its video in to Unity. All other cameras we researched would have been incapable of live streaming video or would have required additional hardware.
In total we are looking at a budget increase of $1250, which will allow us to deliver a fully functional product that completely delivers on the engineering requirements with a polished system. This budget also allows us to handle sudden and unexpected mishaps both covered and uncovered in our risk assessment documents, and will also mean we can iterate and test faster.
For the full document: P19123 Test Validation
Design and Flowcharts
By designing around risk items and taking active measures to minimize risks, we have largely reduced our risks to the 5 risks above.
Plans for next phase
- Utilize subject matter experts early
- Review action items in email with deadline
- Update on any setbacks early on
- Communication on design and progress is vital