Question: what is the next big breakthrough in robotics in the automotive industry?
Many predict that new types of language between a driver and his vehicle will take center stage, making autonomous cars even easier to drive, simpler to control, and more reassuring too, which will not be without displeasure.
More precisely, we are talking here about language interfaces which will allow drivers to give simple voice commands to direct their vehicles, but which will also be able to question them to find out about the elements they encounter along the way, for example, or to collect information that will allow them to avoid traffic jams.
From there to knowing whether drivers will be able to trust them, the question arises.
An innovator
IMAGE TAKEN FROM THE WAYVE SITE
In 2021, Wayve tested and demonstrated that its LINGO-1 technology could be used on the streets of London and four other UK cities, a challenge that was noticed in the automotive industry.
Currently, however, the autonomous driving technology company Wayve is doing everything possible to ensure that the question no longer arises. This company, considered innovative in this area, is a bit like start-up in the development of new autonomous car technologies.
Its software would allow the user to communicate with their vehicle, ask it questions about their driving decisions and receive answers.
The underlying idea of this technology is similar to that of another technology that has been much talked about these years, the famous ChatGPT. Except that in the automotive world, we talk about chatbot (or dialoguer, or conversational agent, it depends).
Currently, there is a certain competition around this type of software to have a tool that offers a natural language, close to that of humans. Software that will give the illusion of thinking for itself and, even better, of being sensible.
To stand out, Wayve therefore merged its old autonomous driving software with a new linguistic model, to create an innovative hybrid model called LINGO-1.
To understand what the car sees and does
The use of natural language is still in its infancy, but Wayve’s technology is changing things right now.
So LINGO-1 synchronizes both video data and driving data (the actions that cars perform, second by second) with descriptions that are heard in natural language that describes exactly what the car sees and does .
In 2021, Wayve tested and demonstrated that its technology could be used on the streets of London and four other UK cities, a challenge that was noticed in the automotive industry.
LINGO-1 appeared impressive. It can answer certain questions, such as: what is the weather like? What dangers do you see? Why did you stop? To which the driver is told that the weather is cloudy, that there is a school on the left or that the light is red…
That said, the curious driver could also ask more general questions: how many floors does the building on my right have?
Focus on safety
However much we think about all the serious and less crucial questions to ask along the way, in the end, the biggest challenge of autonomous driving will always remain safety.
So to create LINGO-1, the designers tested it by asking their team of expert drivers to speak out loud while driving, while explaining what they were doing and why. From there, the company accumulated the data to refine its model, giving it driving tips just as an instructor might train a learner driver.
Consequently, LINGO-1 becomes more reassuring, since the driver has the possibility to ask the autonomous vehicle what it is doing and why it is doing it. Why did he speed up here? Slow down there? Why did you take a break before going around an obstacle? What danger did he sense?
More than one actor
We are talking about Wayve here because it is a pioneer in the use of natural language and its technology is convincing, but other major players use their language model in robotics, notably Google.
There remains the big question of trust. Will we be able to rely on major linguistic models to drive our cars and to be up to date on the information that will be conveyed to us? The question is entirely legitimate and cannot be answered with full confidence at this time.
That said, all the players are working in this direction and efforts are increasing to make the language as precise as possible. Is the chatbot will become the standard in tomorrow’s vehicles? Will artificial intelligence be reliable enough to allow us to drive safely on our roads? This is to be followed closely.