Hacker News new | comments | ask | show | jobs | submit login

It's going to be amusing if GM realizes that Cruise doesn't have much technology and bails on the deal. GM/CMU has successfully demonstrated a Cadillac driving around Washington DC in traffic.[1] Cruise crashed into a parked car in San Francisco.[2]

[1] http://www.cmu.edu/news/stories/archives/2014/june/june24_co... [2] https://www.dmv.ca.gov/portal/wcm/connect/bc21ef62-6e7c-4049...

As someone who used to build the kind of R&D technology that you pointed to at CMU, it is VASTLY different from what Cruise is doing, being so close to customer deployments. In fact, this was brilliant on GM's part, they short-circuited a lot of work trying to convert their fractured list of R&D projects into actual products.

P.S: I'd be worried if there WEREN'T crashes on your record if you're building an autonomous car for me. How much testing did you do?

While I also think that the GM/Cruise acquisition is terrible, it should be noted that the linked accident occurred during "Conventional" mode and not "Autonomous" mode.

Read the report. The automation swerved to the left, then to the right, then the driver took over, but too late. Yes, the vehicle was on manual at the point of the crash.

I refer to this as the "deadly valley". Autonomous vehicles cannot rely on the driver as an immediate backup. This is what happens if a system does that.


Yes it was manually driven at the instant of the accident, but this sort of thing will happen ALL THE TIME during the teething years of SDCs

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact