Everybody Needs a Co-Pilot
How adding contextual AI to car systems will improve driver safety.
Last week, a small R&D team from Neura attended an amazing hackathon hosted by Mazda in Tel Aviv Port. Our goal in the hackathon was to combine Neura’s world — understanding the user’s day-to-day life — and Mazda’s world, which centers on the life inside the car.
Artificial Intelligence (AI) technologies for cars today are focusing on improving driving safety. But they only look at metrics tied to what end-users do while they’re inside the car. They’re blind to critical information about what users do outside the car.
At the hackathon, we had immediate access to amazing data channels from the Mazda cars. During the intensive 36 hours of the hackathon, we managed to build some interesting use cases that provide great insights into the future of car safety.
When It Comes to Car Safety, We Need Improvements Now
Autonomous vehicles are the future for many reasons, the most important of which is safety. Ethics and moral questions aside, I truly believe — and the numbers tell the same story— that a car driving on its own is much safer than having a human driver behind the wheel.
Active safety helps improve a car’s immediate response to dangers, but some dangers could just be prevented in the first place.
KITT (the car from Knight Rider) for the masses is still years to come. In the meantime, we need to make sure cars are safer today. Indeed, companies are already pushing safety technologies to some of their cars.
Tesla’s Autopilot and Mercedes’ new Drive Pilot, which provide basic autonomous capabilities, are amazing feats, but they are reserved for relatively expensive cars. Other car companies are using technology to avoid and reduce the impact of a possible accident using lane correction and automatic breaking. Other advancements in car safety technology include Hyundai’s safety features for its Elantra model, and Volvo’s Driver Alert Control (DAC) which monitors how the car is being handled at high speeds. Tech companies like MobilEye, which has an image processing solution, alert drivers of possible dangers on the road.
Every car company in the world is invested in active safety features, with three- or four-letter acronyms all over the place stating just how a specific model can save your life. But they all focus on response without looking also into prevention.
Car companies know the driver when he or she is in the car. They’re able to measure every metric related to the car itself and some related to the drivers and passengers during a commute.
Unfortunately, no car (or technology) company knows anything about the day-to-day life of their drivers outside the car. This limited access to the driver’s life leads car systems to treat every driver the same, no matter who is behind the wheel.
DUI (of Sleep Deprivation)
The New York Times published an article about research done a few years back reviewing the role driving under the influence of drowsiness plays in fatal crashes. The results are stunning: one in six fatal crashes is caused by tired drivers.
One in six fatal crashes is caused by tired drivers.
There are some solutions that detect fatigue during a drive, but they don’t address the driver before they enter the car. This is the main issue with car technologies today.
The Solution Is Already Here
During the hackathon, we had the opportunity to view the connection between the car and the driver from a few angles. The most appealing one involved car safety, specifically around avoiding sleep deprived driving. Using Neura’s software development kit (SDK) within a mobile app, the mobile phone sensors and connected devices around it were turning into live data channels. By looking at all the channels around a person — where they work, live, exercise, etc. — it gave us great insights about them. We compiled a user timeline that gave us their sleeping profile (one of many other profiles available).
Every morning, when the driver woke up, we could tell exactly how much they slept the night before, and by using other connected devices such as activity trackers, we could even detect the quality of that sleep.
In one example, the end-user, with the Neura-powered Mazda app installed on their device, didn’t sleep very well the night before. It was three hours less than their average sleep time. They had gotten up a few times, the temperature was too hot or too cold at times, and their kid had some bad dreams. When they got out of bed to start the day, a notification would pop up on their phone, suggesting they take the Mazda Shuttle Service to work and avoid traffic hassles while tired.
If the driver disregarded this recommendation, the car was notified of their fatigue status. The car would then turn up its sensor tracking frequency (on a regular basis, this would work on monitoring only), make sure the entertainment system was tuned to upbeat music, and suggest tips to the driver if it sensed them behaving differently from their regular driving profile. These tips varied from opening the windows a little bit to removing the cover from the sunroof to get more light to even pulling over to call someone.
The opportunities for car safety with this kind of technology are almost endless. We soon will see cars that detect road rage, pre-heat on a freezing day so it’s warm enough before the driver leaves to work, and facilitate ride-sharing between a driver and their work colleague who lives nearby, and so on.
Sharing the knowledge between car companies and personalization experts will take the connection between the car and the driver to the next level, creating a true contextual car.
Hat tip to Mazda for showing real innovation by sharing their knowledge with the Israeli start-up community. It gave all of us at Neura and myself a lot of food for thought.
Also, many thanks to Samurai Incubate Inc. in Israel for making this connection happen.
Follow Neura’s new publication Contextual Wisdom to stay updated on the latest in AI, mobile apps, and IoT.
