#MWC18 -Huawei push their AI tech further …. into a car!!

We all know that the dawn of self-driving cars is rapidly approaching, Tesla is pushing the boundaries every day. I have driven the new Nissan Leaf with Pro-Pilot and my standard battlecruiser (aka VW Passat Estate) has got the basics of it in the form of auto cruise control. More and more cars are getting these bits of safety tech built in and they all have one thing in common – serious computing power.

Now I am guessing you are wondering how in the heck this relates to phones and why I am writing about this on Coolsmartphone – a website for phones! Don’t worry the two things are related. The computers that control these self-driving systems have to be intelligent.

Artificial Intelligence (AI) is something we are going to be seeing a lot more of on our phones. They’re now trying to determine how best to help us with our daily lives. I have actually been using an AI supported phone for the last few months and I have, for the most part, been very impressed. I am of course talking about the Huawei Mate 10 Pro. It’s scene recognition mode is stunning and really does help when capturing photos – it takes a lot of the hard work out of getting things set up correctly. I also used it briefly in the stupendously cheap Honor 10 View.

Up until now, these have been the only two devices that support the NPU AI engine that Huawei has created (well, that and the other variants of the Mate 10) – the Kirin 970.

Clearly those guys at Huawei R & D have had a bit too much fun in the past month. They’ve now gone and added this chip into a Porsche Panamera!!

The project name is cool in itself – Roadreader.

Okay, so let me explain a bit more. They haven’t actually replaced the ECU chip with a Kirin 970. Instead they have added a Mate 10 Pro into the car with a whole load of other very techy kit such as robotics and a forward facing video camera in that you’ll see in some of these photos.

Have a look at this infographic to get a better idea of what they have actually done.

Here is the Huawei video which shows the concept off.


We actually managed to get to see this in action today. Huawei let us sit in the car whilst it took control and drove us around upcoming obstacles. These were learnt by the Huawei AI tech and then we told the car how to react, depending on which obstacle it saw.

Here are some of the shots of the cars exterior on a snowy spring day in Barcelona!

Now, some of you may be thinking that there is something different from the Huawei video above to what we actually saw, and you would be right. It isn’t exactly a subtle modification that has been made here. They have added a honking great big video camera onto the roof of the car, however this has all been done just to prove what the Huawei AI tech can actually do, and that big camera is only used for safety reasons. They could actually use the camera on the smartphone itself, but with possible reflections etc it was instead decided to mount a separate one in the roof box.

RoadReader Video camera

The purpose of this camera is to feed the video into the Mate 10 Pro. This is where the actual computation goes on. The live footage is received via a cable into the phone, recognised and processed by the Kirin 970 CPU and the associated NPU. The NPU is capable of processing 30 images per second in real time. The phone will then send the result of the image recognition to an app which will then allow for the robotics in the car to take the appropriate action. In the back of the car, there is a lot of kit to control the Porsche from the phone commands.

So what does this look and feel like for real? Well, Leigh very bravely stepped up into the hot seat to have a ride in the car and was gracious enough to film it along with me getting some outside footage.

I think Leigh summed the experience up very nicely at the end of the video there, it is very cool indeed.

Here are a few more shots of the inside of the car and the UI of the app.

Huawei has stated that this is not something they are actively pursuing – this was really developed to showcase what the NPU on the Kirin 970 and Huawei Mate 10 Pro can do.

That being said, if this was something that they wanted to develop further, the possibilities are endless.

All this from 5 weeks work. What could be done in 5 months, or 5 years?