It's not a function of a color sensor versus a monochrome sensor. It's a function of the dynamic range of the sensor, and I definitely hope they improved this.
@Tam thanks for sharing. That is a disturbing story. I understand auto pilot will disengage when you step on the brake. But isn't there an emergency stop function irrespective of if Autopilot is engaged or not?
The story he's talking about has happened on a few occasions, and Tesla learned from them. In 7.1 the AEB (the emergency stop function) wouldn't activate if you already had your foot on the brake. In 8.0 the AEB changes that behavior to increase stopping power. So you can't accidentally override the AEB by hitting the brake. The throttle still overrides it. In 7.1 the Autopilot radar did not see stopped objects. In 8.0 it has the ability to now see them, but it won't slam on the brakes for them. It might activate FCW, but I don't know if it does. In a future version of 8.X Autopilot it will activate stopping for a stopped car. The AEB system will also activate for a stopped car. This activation hasn't been implemented yet because they have to do the whitelisting to make sure it doesn't falsely activate for road signs, etc. I believe the main purpose of 8.0 is to create the whitelist, and 8.1 will be the full implementation.
I will trust it, when there is sufficient results to provide 100% safety. There will still be occasions where no driver or computer can recover. The new AP should have camera proof that verifies the results. If a deer suddenly darts in front of the car or a meteor lands on the vehicle, there may be no recovery time.
with 360 surround sensors , hopefully the car should be able to make an evasive maneuver when the hapless deer comes darting from nowhere. For the Meteor , we need to wait for the FFS 1.0 upgrade ( Force Field Shield + a ticket to Mars in the glove compartment ) .
The words from Tesla Page: "Full Self-Driving Hardware on All Cars All Tesla vehicles produced in our factory, including Model 3, have the hardware needed for full self-driving capability at a safety level substantially greater than that of a human driver." The safety percentage rate is: "substantially greater than that of a human driver." That means it can fail a lot of time but it can still be "substantially greater than that of a human driver." Elon Musk cited a number for Driverless standard: “But the important thing is that the foundation is laid for cars to be made fully autonomous, at a safety level of at least twice that of a person, maybe better.” If it is 1 times better than human then accidents, injuries or deaths will still happen at the exact same rate. So if Tesla Driverless system is two fold safer than human, does that mean no accidents, injuries or deaths? Could it mean human drivers would result in 30 thousand fatalities annually in the US and Driverless System is twice better than human so the fatalities would be down to only 15 thousand?
Unfortunately Tesla's are not the only cars on the road. We will reach new levels of saftely when all cars can communicate and negoatiate based standards.
Whoa there, 99% is unsafe. As far as US fatalities according to wikipedia it was 1 in 66.7 million miles (1.5 in 100 million) or (1 - 1 / 66.7million) = 99.9999985% safe (from death) with human drivers if AP 2.0 is twice as good then it'll be (1 - 1 / 133.3 million) = 99.99999925% safe (from death)
Wonder if Tesla wants to keep insurance involved so they don't upset yet another highly connected industry.