As part of a PhD research team sponsored by Bentley Motors, DCA has teamed up on an experimental project to better understand how to utilise Artificial Intelligence to create enhanced comfort experiences in the future. The results of this work to date were shared for the first time at the 14th International ACM Conference on Automotive User Interfaces.

This project was initially conceived in 2020 by Mohamed Abd El Ghani as part of an ongoing PhD project sponsored by Bentley, which focuses on understanding the application of AI in an automotive setting. Ravi Jani at Bentley, Mohamed and DCA set about trying to understand if Artificial Intelligence has the potential to drive the next level of in car comfort based on the hypothesis that this will require user trust. “Creating a relationship between the user and AI will be fundamental to its success” commented Mohamed.

The DCA and Bentley teams worked together to experiment on how to create a new era of user comfort, enabled through the use of Artificial Intelligence to identify and act upon users’ needs. "From the outset, we had a very fluid working approach. There were no limitations or restrictions to our ideas, but we knew that we had to physically prototype them, for user trial testing. This is vital in allowing us to understand what resonates with our target users and what doesn’t, so we can focus the direction of the work” explained Dominic Hargreaves from DCA.

Early user trials demonstrated that there definitely needed to be a high level of trust in the AI for users to relinquish responsibility. “You need the ability to tell it that it’s wrong, it needs to understand if you want more or less of its input” Dominic went on to explain.

Alongside this, it was also clear that “Interacting with a car was once a tactile experience, which is on the decline with the rise of car assistants, where the dominant form of interaction is through screen displays and voice recognition. Through this work, we wanted to explore the benefits of reintroducing tactility into the automotive experience, to improve engagement and emotional connection between users and future intelligent cars” said Mohamed.

“This move away from screen- based platforms opened up a wide range of ideas with the potential to create a better connection with the user through harnessing more of the senses” he added.

Together the team designed a physical AI interface that aimed to harmonise the different comfort parameters to create richer in- car experiences. “Imagining it is one thing; to experience it is quite another”, explained Dominic, reflecting the imagine- design- buildtest approach that DCA aspires to work to. Furthermore, building a physical AI interface allowed the team to explore and build on the physical craftsmanship that Bentley has become synonymous with, reinvented for a digital age.

A prototype experience was built, with all the custom hardware and software needed to emulate an intelligent car system. This allowed the team to investigate with users in real- time how they would interact with and control a smart- comfort system to improve user comfort. Part of this work involved the creation of a physical AI interface, which was designed to invite users to interact with it through touch. “Users could use their hands to physically agree or disagree with changes made by the AI comfort system with the interface moving in response, creating a bi- directional interactive symbiosis that re-prioritises tactility,” said Mohamed.

One of the outcomes of the project was clear evidence that positive feedback loops do empower the user, allowing them to seamlessly train and control the AI. “It can be hard to integrate gesture control as part of a user experience. With this in mind, the team focused on the simplest of natural gestures. Increase or decrease, amplify or reduce” said Dominic.

Through user testing, it became clear that there was potential to significantly enhance the user's experience through automated comfort. Traditionally “comfort” is reliant on the user manually inputting changes to control individual parameters that contribute to their comfort. This is the old world where discomfort prompts you to interact and correct the environment. “Based on this research, when artificial intelligence anticipates your needs and adjusts your comfort setting instinctively, we can see that the potential of this new paradigm is pretty powerful” Ravi Jani.

The whole nature of successful AI is to build longstanding relationships with its users. However, even using its learned knowledge of a user’s comfort preferences in particular environments and use case scenarios, AI isn't going to be able to achieve your ideal comfort arrangement instantaneously. It takes time to reduce the interior temperature for instance, if the AI system has not been pre- warned of a user’s journey plans. Not knowing what the AI is doing and that it has the situation in hand, can lead to frustration with the user often manually fighting the AI, which became evident in user trials.

For this reason, the team went on to focus on ways in which the AI can demonstrate what it doing at any particular moment. “We are working in the space where it's making decisions on your behalf, but if there is a problem and it needs to communicate that it's working on it – how will it do this?” explained Dominic. “With a manual dial or control you can turn down the temperature of the air- conditioning and the system works hard to correlate with your latest instruction, eventually getting there. Often the problem with many physical interactions is that there can be a time lag. The feedback is not instant and in our minds, we have made the change but the system isn’t responding”.

How will AI better communicate the common message "I'm working on it” with real credibility. It needs to convey the message that "if you just hold on for a short period, I'll sort this out for you". Alternatively, when some dramatic external influence comes along, how does the car’s AI system let you know that it’s working out what to do? Unfortunately, we have to find ways to overcome users’ natural distrust and scepticism of such interactions, fed on a diet of similar messages delivered up by their personal computers, tablets and phones where “I’m working on it” can often mean “I’m in trouble, and I’m buying some time in the hope that the problem will fix itself.”

The collaborative Bentley and DCA team is building on this work, aiming to further explore our future relationships with AI in an automotive setting. "The intention was to create a conversation that opens up the imagination to the possibilities of how user facing AI will enhance journeys. The ultimate aim of our research is to reduce stress and complexity". Mohamed explained.

“The results so far support an approach that combines traditional physical and groundbreaking digital craft. This is an extraordinary journey to be part of," said Dominic.