Humanising Autonomy’s technology defines how autonomous systems interact with people, making autonomous vehicles safer and more efficient in a range of culturally specific urban environments.
Through the application of AI, Humanising Autonomy has created models that capture the complexity of human behaviour in urban locations around the world, allowing mobility solutions of the future to predict and react to the way people behave in different cultural and physical environments.
As part of Digital Catapult’s Machine Intelligence Garage, this exciting new startup has benefited from access to the expertise, computational power and community support it needed, helping to train complex models, grow its business and explore new use cases and opportunities in adjacent industries.
Teaching technology to anticipate human action
It is said that people are the same the world over, but that’s certainly not true when it comes to the way they behave on the streets – not just in different countries, but also in different towns, cities and neighbourhoods. For example, while jaywalking may be illegal in the USA, in some cities people take much less notice of this law than in others where it might be rigidly observed.
Cultural norms don’t only vary by city, they can also change by district, especially as the behaviour of pedestrians and vulnerable road users is largely shaped by the infrastructure around them. Busy car parks, local markets, school times, road closures, seasonal tourism. There are multiple influences at play all the time, not just from one location to the next, but also from one point in time to another.
While human drivers can swiftly adapt to changing circumstances, autonomous vehicles need to learn how context shapes behaviour, and how to predict and react to the actions of the people around them. Machine learning algorithms need to be trained and validated using huge data sets, which meant that Humanising Autonomy needed access to a significant level of computational power, as well as support for how to best use such a resource. This was their main reason for joining Machine Intelligence Garage as a startup in 2018.
Providing powerful cloud-based services…
Digital Catapult’s focus on supporting early stage startups to overcome hurdles was exactly what Humanising Autonomy was looking for. It joined one of the early cohorts of Machine Intelligence Garage through which Digital Catapult provided access to both computational resources and technical talent to optimise usage of the resources, including data storage, training models and client access via API.
Having credits for access to Google cloud services meant that Humanising Autonomy could train its models using technology that would have otherwise been unaffordable.
… and extending access to human talent.
Although Humanising Autonomy was looking for support in terms of computational power, they found that the benefits of working with Digital Catapult extended much further. As an early startup, being given access to a salary subsidy meant that the co-founders could take the first step towards growing their business by hiring their first employee, a valuable addition to their team.
As a member of the Digital Catapult community, the Humanising Autonomy team can now connect, collaborate and discuss challenges with other startups, a significant benefit that continues over the longer term. And through participation in industry-specific scoping workshops, the team has been able to forge links in manufacturing – a sector which the Digital Catapult team identified as a potential future opportunity for expanding the use and application of Humanising Autonomy’s innovative predictive platform.
Humanising Autonomy recently raised a $5.3 million seed round, and is currently recruiting https://www.humanisingautonomy.com/join-us
Facilitated collaboration with other AI businesses and startups
Accelerated and powered the rate at which predictive models could be trained
Opened doors for new opportunities in manufacturing
Enabled the hire of a new full-time employee (a 20% increase in humanpower)