Hi Just an update on the Uber self-driving autonomous car fatality that I did in a previous video. It's not looking good for Uber as you could have guessed that. all the stories coming out at the moment are that Uber disabled Volvo's SUV safety system before the fatality and well, this is actually not surprising because they have to install their own autonomous system. That's what They're testing their own one with their own what is it? seven cameras, the scanning laser lidar on top, the radar, and you know their own sensor suite of packages that they have to do this so they probably didn't want the cars exist in collision avoidance systems actually interfering with that.

So yeah, it's not surprising that they did that, but or what it sounds like deliberately didn't defeated the collision avoidance system which as we'll see probably could have detected this incident and automatically brake the car if it was a factory Volvo Xc90 that we're looking at here. and the reason that these stories are coming out is because the supplier active of these systems which uses the Intel mobile chipset which we'll take a look at has come out and they want to distance themselves from this saying that hey our system could have detected this and it wasn't used. They actually they must have disabled this and actually Intel themselves which manufacture the chipset here it is. They actually took the very poor quality dark dashcam footage so they took the second-hand s cam footage, fed it through their mobile eye chipset that's used in the factory xc90 which Uber is the car to be used for their autonomous driving cars and also the car used in this incident, and it was able to detect the pedestrian coming across with their walking their bicycle across a second before it happened.

So of course it would have processed that maybe in a like a tenth of a second, say 0.1 seconds. That probably would have had almost a second to decelerate. And of course, if that was the case, they they couldn't have probably prevented the accident. the car couldn't have stopped in time, it wouldn't have had the distance to do that from what is it, 43 miles an hour or something that the car was actually traveling at, But hey, it would have lessened the impact.

And and that's from just the second-hand really dark dashcam footage. as we'll see later on in this video. so they're saying look, our system would have performed much better if it was a factory. Volvo And it sounds like Uber have deliberately disabled that so that could get them into a lot of trouble.

Anyway, let's have a look at the Volvo Xc90, which is this is the 2017 brochure, but as we'll see, the technology's been around since like 2013 I Think and it's used in lots of cars. Volvo Aren't the only one that have at the Xc90. Lots of cars on the market, even your cheaper ones now have these collision avoidance detection systems not only Lane Guidance but detection avoidance as well. Here it is, it's looking out for you.

City Safety blah blah blah blah blah. City safety can identify other vehicles, pedestrians, cyclists, and large animals in your path. it won't. It warns you of any hazards and if necessary, will brake automatically to help avoid and mitigate collision.

So there you go. It's built in to the car and even like older model cars, it's got exactly the same you know 2016 brochure and going back. and if we have a look at mobile I hear it's now an Intel company and I've actually done a teardown video of one that was used in a high owned I I'm not sure which Hyundai it is, but it uses the the older technology are the Q2 chipset and you can see the camera Here it's got is just a building camera with a very high data rate through to the chipset because this thing does like I think the new chipset. the Q3 does 0.25 trillion operations per second so it's really amazing.

I'll show you in a second what it can do so this has generally mounted up in the front of the car. I'm not sure of the exact details of the Xc90 used in the Uber here used by Uber here, but it would have had a similar camera and they must have disabled this module which could have detected and automatically braked for pedestrians and things like that. So anyway, it's very impressive what this technology can do, So we'll just roll some footage here and of Coretech Ssin tracking of all the pedestrians, cyclists, people walking across cars in front and it can do lanes. Of course this is daytime so obviously it's going to perform really well.

but you can see how advanced this is and this is not new and if we have a look here you can see that. the one that I tore down the Iq2 chipset that's actually from 2010 but the Q3 chipset dating from 2014 that's used in lots of cars on the market. Not only the Volvo it's got autonomous braking system for our pedestrian, even does animal detection and things like that holistic path planning Road Lane reconstruction and all sorts of collision avoidance. so it's not.

and if you actually go and look at, some of the videos are actually quite old here look four years ago. Volvo Collision avoidance systems demonstrating. here you go, there's mobile ice tough dating pedestrian collision warning systems dating back to 2012. This is not new technology and at night time as well you might think it's just daytime, but here is the mobile Eye chipset.

At night it can detect all these cars and can detect the lanes and everything else and this is really quite low light footage. I'm not sure what you know camera is if that's the actual camera used in the car or the weather not that's a secondary and they're just processing a bit. you know even at nighttime it's still very impressive and it's amazing what it can do in crowds here. Just check it out.

It's detecting all these people and crossings and all sorts it like it's Fanta It detects arrows on the road. It's like practically just built into this mobile eye chipset which are started out as just a Lane Guidance type warning thing. He's incredibly advanced now with its point 2 5 trillion operations per second and it cannot look and there's a zebra cross in and there's people and it can do. It can do most things that an autonomous car can do with just the camera.

It's really amazing and Volvo made the news back in 2015 in that their system couldn't actually detect kangaroos here in Australia hopping across the road and it's a huge problem here in Australia Seriously like especially at dawn or dusk when because kangaroos are typically nocturnal animals. they come out in the Twilight and it's a night and they hop across the road and you can't see them Anyway, their software wasn't designed to detect the hopping of the kangaroo. it actually confused them. So they actually came to Australia in their Xc90 and they actually look updated their software.

You can see it actually detects the kangaroos hopping across the road and they updated their software to actually you know and to get this anomaly just here for Australian condition. So there they're really on top of this stuff And that was back in 2015. So yeah. anyway Volvo haven't commented on this yet and probably for you know, sensitive legal reasons I guess.

But yeah Mobile I, Intel, Mobile Eye and active who supply the sensors have come out and said if you using the crap dash cam footage we could have detected that Now of course you saw the footage in the previous our video and warning I will actually show it again here. So it does show someone getting hit by this autonomous uber and but you saw that footage was really dark and it was almost. It appeared as though they come out of the shadows. but of course that has to do with the type of sensor used, its exposure conditions and and contrast and all sorts of you know other things.

So there are quite a few people who actually went out and went to the same location where this accident happened at night and actually shot footage of what it's actually like. but this youtuber here actually I shot this footage in 2015. It just happened to be the exact same location where it happened and this was shot with a stereoscopic camera at the top of the Sun roof of the car. So let's actually roll that you can see the bridge here.

this is the bridge that it's just past this bridge where it happened and you know see my video a previous video for look but you like can see here's the dark shadow that the pedestrian happened right about here. but you can see that there was still plenty of light there to actually see this so it shouldn't have been a big problem. And here's somebody who strike engine car TV actually went and shot footage what looks like on a mobile phone and with the actual footage and have warned in here it is and the cyclist will come out any second here and they sync that up and you can see that even with the crap sensor in what looks like presumably are like a mobile phone, it's really easily able to see that. So yeah, the footage provided by that dashcam from Uber is I don't know, isn't like it's some old technology I Don't know what's going on there but that is really very dark.

that footage. it really is quite remarkable. When you saw it it's like wow look look how dark that is. It's amazing.

whereas the actual visibility and these are time synced apparently huge difference. So yeah, not sure what's going on there and here's another one coming up to the bridge it looks like this is shot with a mobile phone as well, just hand held by the looks of it. let's have a look and you can see look in Ian's pedestrian so that here's the shadow happened right here. so like there seems to be no shortage of light there.

Yes there is a dark shadow that just happen to be in that spot, but geez, if a tiny sensor in a phone can detect that, Yeah, it's not looking great for Uber, is it? So there you have it. Just want to share the update on that and how lots of cars on the market have this system that probably could have detected this just with visual cameras, but the Uber has the scanning laser lidar on the top just spinning around. Update Rates: We've had comments on the previous V from people who've worked on those systems and they actually give the resolution of that lidar. And from memory I Think they said like three, two or three centimeters resolution or something like that.

Updating at least ten times the second. Don't quote me on that, but yeah, it should have been able to detect it. just the lidar alone, let alone the radar and let alone the seven cameras that they got on there don't know how many are pointed forward. probably at least a couple could have detected it as well or should have.

So yeah, it ain't looking good for Uber Anyway, leave your comments and thoughts down below and I'll link in my the tear down video of that type mobile eye sensor the older technology one at the end. Check it out. Okay, see you next time.