Aerospace Xelerated: themes for Cohort 3 (pt. 1)
Assured autonomy, autonomous navigation, and generative design are the first three themes for the upcoming Aerospace Xelerated programme.
tl;dr — 🛫 Applications for Aerospace Xelerated accelerator programme are now open, and close 8th October 2021. We’re looking for world-class startups at seed to series A stage that are building (software-focused) solutions in the Autonomy and/or AI fields. Find out more, apply online, or book an Office Hours call.
In the upcoming Autonomy and/or AI-focused cohort, there are a number of specific sub-themes where we’re looking for world-class startups to support. Over the next two blog posts, we will be illustrating what Boeing and its industry partners are looking for from startups focusing on these themes. This article covers Assured Autonomy, Autonomous Navigation, and Generative Design. In pt. 2, we will expand on Smart Maintenance, Adaptive Learning, and Reduced Workload.
A side note: our definition of aerospace revolves around the design and manufacture of aeroplanes and the peripherals around that process; rather than spacecraft or satellites. Startups with hardware-only solutions are unfortunately out of scope for this programme. We may hold future programmes with a stronger hardware focus so do stay tuned.
Assured Autonomy
“The true value of autonomy lies in its ability to stay on mission longer, cover more areas and provide more immediate, actionable knowledge.”
Autonomous systems can accomplish tasks independently in complex and unpredictable environments with minimal supervision from human operators. Their applications vary widely, ranging from defence operations, aerial refuelling, and harsh environment explorations, amongst others. The true value of autonomy lies in its ability to stay on mission longer, cover more areas and provide more immediate, actionable knowledge — prioritising humans for the most important actions and decisions.
Despite tremendous advances in constructing autonomous systems over the past decade, challenges remain that impede the deployment and adoption of these systems in applications where safety is critical.
We are particularly interested in seeing startups in the assured autonomy field that are tackling the following challenges faced by the industry:
- Existing autonomous systems are low-level and still require substantial human involvement, limiting the benefits of autonomous systems. It also introduces critical challenges in ensuring efficient and effective interactions between humans and machines. This includes questions like: how can systems ensure the operator reacts to things that need their attention effectively?
- To achieve higher levels of autonomy that can be applied to uncertain, unstructured, and dynamic environments, more data-driven machine learning techniques are required. There are many open systems science and systems engineering challenges that need to be tackled whilst developing and utilising these advanced techniques.
- Existing machine learning techniques are inherently unpredictable and lack the necessary mathematical framework to assure correctness. This limits applicability of autonomous systems to operations that require predictable behaviour and strong assurance.
“Challenges remain that impede the deployment and adoption of these systems in applications where safety is critical.”
Whilst a majority of assured autonomy interests lie in applications related to defence, explorations, and intelligence, drones also have a growing commercial presence. Industry members are increasingly looking towards offering unmanned systems as a service product. A key consideration towards this industry shift is the need to ensure commercial autonomous vehicles are still used in a safe and controlled manner. Hence, we are also interested in startups with software solutions in this area.
To read more about Assured Autonomy, our Boeing experts recommend DARPA’s resource.
Autonomous Navigation
The second challenge we want to see solutions for is related to Autonomous Navigation. To define this, we are referring to technologies enabling vehicles to plan their path and execute their plan with zero human intervention.
“Most algorithms have only been demonstrated to successfully operate in well-structured and highly predictable environments.”
Autonomous navigation has made rapid advances through agile develop-test-develop-test approaches and/or simulation-based approaches. Both methods have accumulated data to help train and tune algorithms. As a result, these algorithms have only been demonstrated to successfully operate in well-structured and highly predictable environments, meaning deploying unmanned aircraft in complex environments would require at least one human operator. The challenge of ensuring independent navigation of UAVs, in particular, involves the requirement of tracking an additional z coordinate to map out more complex aircraft orientations such as roll, pitch, and yaw.
One of the leading solutions in this area is Boeing’s ICOMC2, a software that can control multiple unmanned aircraft with one ground control station, which still requires intervention from an operator.
We spoke to Peter Wright-Gardner, a Senior Systems Engineer at Boeing, on what’s required in future autonomous navigation solutions:
“To fully leverage the benefits of autonomous vehicles, autonomous algorithms need to take the human out of the loop and give the platform the ability to ‘think’ for itself. The big challenge once this is achieved will be convincing regulatory authorities.”
The path to achieving complete autonomous navigation is multi-factored. It requires advances in technology in remote sensing, tracking, information processing, data integration, and flight operations and training.
Thus, we are particularly looking forward to learning about solutions that tackle any one or more of these challenges that can bring true autonomous navigation to unmanned aerial vehicles.
Generative design
The 3rd theme we are discussing today is generative design.
“Generative design can provide major benefits and advances in tackling key engineering, architectural, and systems challenges.”
Generative design is an exploration process to create design alternatives from a single idea. In theory, generative design software explores all possible permutations of a solution, tests and learns from each iteration to identify the best solutions that suit defined design goals.
Within the aerospace industry, generative design can provide major benefits and advances in tackling key engineering, architectural, and systems challenges. This can accelerate the process of reducing aircraft weight significantly so less fuel is consumed and hence reducing the industry’s environmental impact.
In particular, generative design has been used to rethink the design of cabin partitions, vertical tail planes (VTP, or vertical stabilisers), and even aircraft manufacturing factories to optimise workflow. As a relatively new technology involving complex and highly automated algorithms for evolutionary design and topological optimisation, there are few generative design software designed specifically for aerospace use-cases.
Human designers still have a central role in creating and defining the most suitable design solution. To serve as an auxiliary tool, generative design software needs to grant designers the ability to a) intuitively navigate through the whole design space, b) update parameters at each step of the iterative process, c) choose from a set of different product configurations rather than a singular solution.
It is important to highlight the tight relationship between generative design and additive manufacturing. Many generative design alternatives involve complex geometries like lattice structures and are difficult to build using traditional manufacturing methods. By combining generative design with additive manufacturing, it becomes possible to build complex and multi-functional designs by using a mix of different materials so that different properties are distributed in different zones of the same part.
However, additive manufacturing processes and markets are still maturing. There is often a range of shifting variables and material requirements introducing uncertainty into the manufacturing process, hindering the translation of design to manufactured parts. We would be particularly interested in generative design solutions that take these moving parts into account. Additive manufacturing is a theme on its own for this cohort, so we will cover it in much more detail in a separate article.
Have you developed a solution in the AI and/or Autonomy theme? To apply for this cohort, head over to our website.
Learn more about Aerospace Xelerated in our FAQ or watch the recap of our Ask Us Anything webinar. You can also book an Office Hours call to discuss your queries with the programme team.
Stay up-to-date with our latest updates via Twitter, LinkedIn or by subscribing to our mailing list.
