Mobileye's Responsibility-Sensitive Safety Mathematical Model Explained

This 20-minute video explains Mobileye's safety model for autonomous vehicles and the formulas used by an AV to make safe driving decisions. The video shows how each of the 5 rules of RSS work using real-world driving scenarios.
0:00 - RSS Introduction
2:18 - Rule 1 Do not hit someone from behind.
5:11 - Rule 2 Do not cut in recklessly.
7:54 - Rule 3 Right of way is given, not taken.
10:48 - Proper Response
12:34 - Rule 4a Be cautious in areas with limited visibility.
14:14 - Rule 4b
16:18 - Rule 5 If you can avoid a crash without causing another one, you must.
17:55 - Summary
About Mobileye:
Mobileye is leading the mobility revolution with its autonomous-driving and driver-assist technologies, harnessing world-renowned expertise in computer vision, machine learning, mapping, and data analysis.
Our technology enables self-driving vehicles and mobility solutions, powers industry-leading advanced driver-assistance systems, and delivers valuable intelligence to optimize mobility infrastructure. Mobileye pioneered such groundbreaking technologies as True Redundancy™ sensing, REM™ crowdsourced mapping, and Responsibility Sensitive Safety (RSS) technologies that are driving the ADAS and AV fields towards the future of mobility.
Connect with Mobileye:
Visit the Mobileye WEBSITE: www.mobileye.com/
Like Mobileye on FACEBOOK: / mobileye
Follow Mobileye on TWITTER: / mobileye
Join Mobileye on LINKEDIN: / mobileye

[@EinzigfreierName]

Looks much more feasible to me compared to a purely AI driven approach as the behaviour of the system is much easier to predict, design, control, and verify. This should also make certification for some levels of autonomy easier.

[@GreatRaphael]

Brilliant !

[@MaddoxBrian79]

So excited for this!!!

[@eyal.herlin]

Excellent video! In the "Rule #2 Do not cut in recklessly" segment, why would you assume a human is driving the car? Wouldn't an EV with a faster response time be the worst case scenario, leaving you with less time to react?

[@DiapaYY]

Isn't the worst case if the other car reacts slowly? If it reacted faster it could more easily avoid a collision. That's also why they're also using the other car's minimum reasonable breaking capability instead of its maximum reasonable breaking capability.

[@eyal.herlin]

@@DiapaYY I will have to rewatch then. My thinking was that the quicker a change around you happens the less time (by definition) you have to react.

[@eyal.herlin]

@@DiapaYY You were right. I missed the part about them talking about the other car trying to avoid the collision too (and not just some general change like i was thinking). It's at 6:19: "here again we need to make an assumption about the minimum lateral breaking of the white vehicle. in other words how fast can it steer away from a lateral conflict".

[@albertwang5974]

The core of safety: keep a certain spacetime ownership unique!

[@michaelwoon5242]

5:19 is that good driving behavior? Wouldn't a normal driver see the car ahead is pulling out of the parking spot and wait for them to clear before passing? The Ego gets in a pretty risky situation with the on-coming car approaching and nowhere to go. Good thing the driver took over. Some companies are taking safety drivers out and putting the public at risk.

[@Extys]

Mobileye is the way high-IQ people solve the autonomous driving problem.

[@Factoryseconds123]

Nah that's Tesla 😂

[@cyberslim7955]

@@Factoryseconds123 That level 2 fake junk? Stop lying to yourself and the rest of the world!

[@coez9373]

I think rule #2 should be "people on mopeds playing stu*id games, should be allowed to win stu*id prices"

[@willijiem]

It's like the Three Laws of Robotics by Asimov is coming to the real world!