Name: Amnon Shashua和Mobileye團隊--改善道路安全的車輛願景 (Amnon Shashua and the Mobileye Team - Vision for vehicles to improve road safety)
Uploaded: 2021-01-14T11:22:57.000Z
Duration: 6 min 59 s
Description: 【看影片學英語】數萬部 YouTube 影片，搭配英漢字典即點即查，輕鬆掌握單字發音與用法，長久累積看電影不必再看字幕。

基礎被動語態

90 percent of traffic accidents are caused by human error

And the biggest risk factor in road traffic - is us.

Forward collision accidents, lane departures.

Our dream is to eliminate all those accidents totally – we call that “Vision Zero”.

Technology from the holy city of Jerusalem, with a noble objective:

Cars that won´t crash. This basically takes human error out of the equation.

We should be able to save tens of thousands of lives with Vision Zero...

Here at Mobileye, Amnon Shashua and his team are working on technologies that are transformative.

It all begins with the understanding that just like a human can understand, by looking,

what things are, where they are, and infer what they intend to do,

the computer should be able to do the same.

Making a computer see and interpret the world is game changing.

Back in the late 90s when Mobileye was founded, we understood that such technology can make a life-saving impact.

There was clear interest from the industry for Driving Assistance Systems -

but the solutions were expensive and cumbersome.

The industry thought that they needed stereo cameras they needed radars, they needed laser range finders.

All those cumbersome systems. In fact, the human visual system

relies on a diverse set of cues, all of them are monocular in nature,

like perspective, motion, shading and context.

Although it sounds unintuitive, doing this with a single camera, and mimicking the processes

of the human visual system not only make it less expensive but also make it work better

than other systems like stereo system relying mostly on geometric triangulation".

Professor Amnon Shashua of the Hebrew University in Jerusalem returned from MIT

where he completed his studies in computer vision and artificial intelligence

with a vision to create a computerized “eye” for the sake of road safety.

After founding the company with Ziv Aviram, he thought that the perfect person to lead the R&D

of the nascent company would be his former student Dr. Gideon Stein.

Shashua gathered other team members with diverse technical skills to solve the many different challenges.

The first challenge is to detect the vehicles. You start by putting a bounding box

on the vehicle in the image which is natural for a camera. Then comes the measurement.

That´s the bigger problem because a camera is not a natural distance measurer.

We had to develop smart algorithms to solve this problem.

Prof Shashua's team at Mobileye have the objective to integrate

a cheap and simple driver assistance system into any vehicle.

So this is the basic system. So we have here the camera sensor and the lens, so this is the camera part.

Here's the processor, our EyeQ processor. And this is the board which does everything which we need.

The forward-looking camera behind the rear-view mirror sees what the driver sees.

The software interprets the camera image.

Artificial intelligence now has to draw the correct conclusions.

But how can that be done without precise sensor data?

Instead of using precise range measurements, the algorithms use the information from the camera's image

to make a clever estimate. For instance, when calculating the distance.

There we use perspective. We look at the bottom of the target whether it is a vehicle or a pedestrian,

where the bottom is in the image. And its distance from the horizon gives us information about the distance

to that vehicle. It's a bit more complicated than that, but that's the basic idea.”

Mobileye has also found ways of making it work at low light, in rain, and on bumpy road surfaces.

We were really formed in a garage but through a series of inventions

we were ahead of everyone else. We filed patents, we improved the technology

To detect an impending collision in the camera image, Mobileye´s system uses a further trick:

The second application is working out the time to collision. When are we going to hit the target?

And there, we look at the expansion of the target in the image

and that tells us something about when we're going to collide.”

The objective: to warn the driver - in good time - of an impending crash.

Today, Mobileye's “smart” driver assistance system is a world leader.

In 2017 the company was bought for more than 13,5 billions Euros by the chip giant INTEL

and Mobileye technology is now found in over 40 million vehicles.

In this car, Mobileye installed a system with special image output -

- to demonstrate everything the computer eye can see and what it calculates.

During the car journey through Jerusalem, the screen shows the driver assistance system at work.

So what we're doing is, we're detecting all the vehicles in the scene

and for each vehicle we put the bounding box around the vehicle

Every 27 milliseconds, algorithms calculate the distance and relative speed of the objects in the field of view.

And the one vehicle which we care about most is the CIPV, the Current In-Path Vehicle,

and that is the vehicle from which we want to keep a safe distance.

If it's an adaptive cruise control it'll monitor that distance,

or just make sure we have a two-second headway from that vehicle.

Meanwhile, a whole family of products is based on Mobileye's developments.

If danger threatens, the systems warn the driver visually and acoustically -

and can even be engineered to trigger an emergency braking manoeuvre all by themselves.

The driver assistance system research also gave rise to another project:

Mobileye is testing a fleet of driverless cars.

Over the last seven years, the number of European patent applications

in the driverless car sector has risen by 330 percent.

Mobileye is at the forefront here as well.

Driver assistance systems are still the project closest to the hearts of the inventors at Mobileye