The Recent Accident Concern and many in last year pose question weather self-driving systems can reliably avoid accidents. Last year I came across this article addressing Ehical AI issue The infamous ““trolley problem”

Many hard thing to pin down with conventional deep learning ( a method that uses layered machine-learning algorithms to extract structured information from massive data sets). A hard time even generalizing across different frames of a video, labeling the same object on minor shifts can completely change the system’s judgment.

One of the big issue always was Lane change detection. In the last few weeks, some drivers have noticed that Autopilot seems to be detecting lane change intention from other cars, then slowing down to let other drivers change lanes in front of it.

Good to see that Tesla is continuing to improve Autopilot’s abilities, last 2 weeks system has gained two new capabilities:

-Detecting stop lights
-Detecting lane change intention from other cars.

Watch this Test drive video just uploaded on this Reddit Video https://reddit.app.link/sbrKojNWiV.

which is showing Telstra model 3’s display and dashcam footage side-by-side which indicates that Autopilot saw a car intending to change lanes then applied brakes to make room for said car. Model 3 is on software 2019.5.15, which started rolling out about two weeks ago.

A walk in the past, Ever since 2015, from Elon Musk who a fully autonomous Tesla by 2018 to Google. 2019 predicted Delphi and MobileEye’s Level 4 system or Nutonomy plans thousands of driverless taxis on the streets of Singapore or. GMs fully autonomous car 2019.

Looking at today’s news that Hackers crack Tesla Model 3 in competition, Tesla gives them the car, I feel Tesla and other OEM manufacturers shall not hurry to meet their commited deadline and focus on mitigating all risks in first place.