Clear Prop! | #3
DALL-E 2 & AAM, Agile Vehicle Development, Wireless Connectivity & Localization for Urban Ops, Coaxial Rotors & more
Very rarely in the history of innovation do hype and product utility meet at the same moment in time. These past months have demonstrated one such rarity taking the world by storm, perhaps breaking early adoption records on Roger’s Diffusion of Innovation. Of course, here I am talking about the AI/ML-based text-to-image tools popularized by OpenAI DALL-E 2, and more recently by Midjourney and Stability AI.
For those of you who may not be familiar with these incredible tools, they essentially let you input a descriptive sentence, which is then transformed - quite magically - into a plausible and sometimes highly realistic image. While the existing tools are not perfect, with a few little tweaks to the prompt, they work pretty well most of the time.
Just to give you a gist of what’s available to consumer today, here are a few examples I’ve been playing around with using DALL-E 2:
If your jaw isn’t dropping by the sight of these images and think this is just a toy, you are missing the point. These tools are already fueling an outburst of creativity among creatives and non-creatives alike. From automated & hyper individualized marketing campaigns to swiftly preparing images for business presentations, and AI-augmented art exhibitions to album artworks, we are already seeing AI-generated images creep into numerous use cases.
The question is, where is this tech likely to impact Advanced Air Mobility? In a highly regulated, precision-based industry where the adoption of ML-based tech is slow, there still are meaningful opportunities for DALL-E-like tools across the value chain:
Rapid product conceptualization - whether you are a product designer or an engineer, being able to quickly visualize an idea and iterate over variations is key to conceptual & preliminary design. An example in AAM design may be quickly iterating over various in-cabin designs for optimal passenger experience.
Patent visualization - being able to visualize existing patents brought together within novel ideas could be an interesting way to generate highly original yet patent-conforming products on the get-go. For example, visualizing a biometric security screening system at vertiports with clear sub-system patent identification may not only accelerate the preliminary design process but future-proof the product for regulatory compliance.
Company branding - whether presenting a futuristic vision at a corporate shareholder event or a startup simply looking to generate logos in a quick & lean manner, AI image generation can support firms in their branding, PR, marketing, advertising, and the like. For example, an eVTOL operator looking to launch in some city could quickly show how a UAM network may look like to its residents with impressive. futuristic, and beautiful graphics.
A couple of years into the future when we see next-gen tools for automated video, 3D image, and perhaps whole generation of scenarios & workflows, we’ll increasingly see engineering, product design, and business development being augmented by these tools.
But enough of AI. Let’s dive into this edition’s papers.
#1: Highly Iterative Concept Development for Requirement-Compliant Air Taxis
Researchers from RWTH Aachen and Fraunhofer Institute for Production Technology IPT propose an iterative aircraft development process that is aligned with the uncertain regulatory environment and fluid economic demand of AAM. Various methodologies are compared & contrasted across qualitative criteria that are deemed valuable from the conceptual all the way through detailed design. By scoring each approach through expert interviews, a merge between a double diamond model based on design thinking and SCRUM-based agile development model is proposed.
Key takeaways:
Mission/aircraft requirements and conceptual design is executed through a double diamond model. This involves an open-ended problem definition and solution finding process that is user-centric, without prematurely jumping into product concepts before intricately honing down on the problem.
SCRUM, a project management approach used mainly in software development, is applied for the preliminary design phase where the conceptual design is transformed into more concrete system design choices. It’s a highly iterative approach based on 2-4 week sprints and frequent standups - “daily scrums”. Such an approach may support teams in avoiding unnecessary and often costly rabbit holes, which are better eliminated early on before the regulator is engaged for certification.
The authors propose this methodology based on expert interviews and thus, it will need to be tested out in the field with regards to its utility and effectiveness.
The BFD: There is a constant tug-of-war between iterating on engineering & product decisions and freezing design choices, especially for eVTOL. Designing a vehicle that concurrently meets market demand that is yet uncertain, certification standards that continuously evolve, and tech that advances is no easy feat. Therefore, any end-to-end development methodology that is able to account for such dynamic environments is key to success in AAM. On the other hand, incorporation of user-centric approaches such as design thinking can lead to the accommodation of extreme users including those with disabilities. Check out EmbraerX’s exceptional work on this here.
#2: Wireless Connectivity and Localization for Advanced Air Mobility Services
In a joint paper by North Carolina State University, University of South Carolina, and University of North Texas, researchers tackle the problem of providing reliable and robust connectivity & localization services to AAM operations. Aerial vehicles with automation will require sub-100 ms latency and minimal disruption caused by base station handovers. Currently in many places, the cellular ground infrastructure is insufficient for these purposes. Therefore, researchers simulate various locations and configurations of base stations, determining their effect on coverage, handovers, and localization performance.
Key takeaways:
Existing cellular networks present 3 main challenges for AAM: 1) base station antennas are tilted toward the ground which is optimized for terrestrial users, 2) aerial vehicles would be within line-of-sight of several base stations, causing strong interference, and 3) interference caused by ground reflection due to the down-tilted base stations.
Coverage can be maximized by introducing uptilted antennas, optimizing their angles, location, and the altitude of the skylanes. Minimizing handovers can be achieved by assigning an optimum band of speeds and altitudes for the vehicles flying. Except for low altitudes, localization errors can be minimized by maximizing the total 3D area created by the base stations and the location of the vehicle.
Novel technologies such as intelligent reflection surfaces (IRS) to increase coverage, digital beamforming for better handovers, and “crowd-sourced” sensor data obtained from other aerial vehicles for localization are promising ways to further address the problem.
The BFD: High bandwidth, low latency, and reliable communication are key to ensuring safe and efficient AAM operations both in urban and rural areas. For Unmanned Traffic Management (UTM) to manage the airspace effectively, all current and future airspace participants must have a common operating picture about traffic, schedules, off-nominal conditions, and the like in real-time. However, especially in dense urban canyons, sufficient cellular coverage within AAM altitudes (up to ~2000 ft) currently is not provided. Thus, there is a large market opportunity here for telecommunication companies to roll-out infrastructure and bring significant value to vehicle & vertiport operators, and UTM service suppliers, alike.
#3: Evaluation of Heave Disturbance Rejection and Control Response Criteria on the Handling Qualities Evaluation of Urban Air Mobility (UAM) eVTOL Quadrotors Using the Vertical Motion Simulator
Researchers from NASA Ames compare the handling qualities of a six-passenger quadrotor eVTOL concept across various vehicle control designs. Utilizing a vertical motion simulator, 1 collective-controlled and 3 rotor speed (RPM) controlled vehicle variants were test flown by pilots across 2 main performance standards: 1) ADS-33 (used for military rotorcraft) and 2) a tailor-made one for UAM. Multiple scoring methods were used to convert the qualitative feedback of the pilots into quantitative ones, including the Cooper-Harper Handling Qualities Rating, Bedford Workload , and the Pilot Induced Oscillation Rating scales.
Key takeaways:
Overall, the eVTOL variant that was controlled by collective was found to have the highest handling quality ratings for most scenarios.
Of the 3 RPM controlled eVTOL variants (each having been tuned to different flight control parameters), the variant that was tuned to have a lower heave response time performed the best in hover-based maneuvers. For other maneuvers, the handling qualities differed across variants.
UAM performance standards - as opposed to ADS-33 - for certain mission task elements are more stringent, albeit more time is given for pilots to execute those maneuvers. Therefore, handling qualities ratings worsened for certain UAM scenarios.
The BFD: Traditionally, rotorcraft vary the pitch angle of the blades to maneuver in air. However, primarily due to increased weight concerns, most eVTOL designers have opted for variable RPM rotors to do the same. Thus, comparing & contrasting both approaches across the same handling qualities performance standards can inform regulators from a certification standpoint. Given that we are still a few years out from the first vehicles being certified, research that supports regulators and aircraft OEMs on certification standards is key to achieving that goal.
#4: Rotor Performance Predictions for UAM - Single vs Coaxial Rigid Rotors
In this paper from Pennsylvania State University, researchers propose a computationally efficient computational fluid dynamics (CFD) tool to predict the performance of single and coaxial rigid rotors. Stacked rotors can enable high-speed flight and the smart phasing of rotor angles can introduce noise reduction, as demonstrated in research by the US Army. However, a computational tool that can predict thrust and torque under various flight conditions is key to optimizing coaxial rotors for eVTOL applications.
Key takeaways:
When building the CFD model, the fully-turbulent airfoil assumptions generated a much better prediction capability as measured by experimental results, compared to the free-transition airfoil assumptions. For example, the predicted thrust and torque using the former assumptions were within 1.4% - 3% of experimental values.
For most flight conditions, the upper rotor thrust and torque in the coaxial configuration can be approximated well with only the single rotor modeled. This is especially true for high induced velocities as compared to the flight speed.
During high descent rate conditions, the presence of the lower rotor in the coaxial configuration delays the effects of vortex ring state on the upper rotor.
The BFD: Running CFD models can sometimes take several weeks to accurately predict the performance of rotors, propellers, and the whole airframe itself. Thus, any reduction in computational cost with acceptable decrease in fidelity is highly sought after across the aircraft development cycle. Several eVTOL designs already incorporate stacked rotors including EHang 216, Urban Aeronautics CityHawk and LIFT Aircraft HEXA. Time will tell which eVTOL propulsion choices are the most efficient, safe and quiet.
#5: Investigation of Merge Assist Policies to Improve Safety of Drone Traffic in a Constrained Urban Airspace
Delft University of Technology researchers propose a one-way airspace structure for small UAS operations over grid-based cities. Modelling the air corridors exactly over streets and avenues, New York City is utilized as a use case to analyze the effect of different merging strategies as drones transition from one corridor to the next. Delay-based and speed-based merge-assist strategies are simulated. Metrics such as the total number of pairwise conflicts & loss of separation, distance to the nearest drone within the same altitude layer per street, and traffic accumulation per street are quantified as proxies for the effectiveness of the merge-assist strategies.
Key takeaways:
Airspace above streets are vertically divided into layers comprising corridors for cruise and dedicated layers for merging, with additional vertically-stacked layers based on a pre-set direction of traffic.
Speed-based merging - i.e. adjusting the speed of the merging drone rather than fully stopping (delay-based merging) - is somewhat more effective in reducing conflicts and loss of separation events.
The topology of the urban grid i.e. the length of streets and the number of turns associated with flight plans is significant in the effectiveness of the merge-assist strategies.
The BFD: While the ideal structuring of airspace would be without corridors and drones being routed freely across space & time, the physical constraints set by dense cities such as NYC necessitate more complex airspace design and thus merging/de-merging approaches. Especially during the initial phases of AAM deployment over cities, dedicated air corridors will ensure that vehicles do not lose separation and fatally collide. While this research does not assume advanced UTM services such as intelligent scheduling, sequencing & spacing nor demand-capacity balancing, understanding optimal merging characteristics is key for safe urban ops.
Before I end this edition, I want to give a mighty shout-out to the Allplane Podcast - a great resource that I have in my arsenal where I acquire expert insights into innovation & sustainability in aviation. Check this out to hear from visionaries & entrepreneurs that are working to redefine the future of air mobility!
This edition’s sponsor is Vertical Flight Society. VFS is an amazing resource that I use to access leading technical papers and workshop decks for my work. Whether you are an investor, engineer, researcher, or an entrepreneur, becoming a VFS member can give you an asymmetric advantage in the AAM space.
Membership gives discounted access to more than 15,000 technical papers, presentations and articles. Learn more here.
Great job Pamir.
Thanks Pamir.