I saw this post on Hacker News today: Uber Wasted $2.5B on Self-Driving Cars. This is no surprise given Anthony Levandowski’s statement about the inefficacy of LIDAR. Uber, with what they’ve attempted to build thus far, has no chance in autonomous driving. Tasha Keeney (ARK Invest) believes that autonomous ride hailing will dominate the ride hailing industry and will account for the vast majority of market share in the space considering how much lower robo-taxis would cost.
Tesla is way ahead of the game, with the largest collection of data in terms of miles driven. Assuming that full autonomy can be achieved, Tesla is in the best position to achieve it. On Battery Day, Elon Musk said that the autopilot rewrite would be available to beta testers in about a month, and reaffirmed that the new approach will work. Either he’s bluffing, genuinely mistaken, or in fact correct that it will work (at some point in the very near future, whether a few months or a few years from now). Many believe full autonomy is not even possible. I tend to think it is possible, however I’m not familiar with the biggest challenges, other than the fact that there are so many edge cases, ones you can’t think of and specifically train for, and that a general AI would probably be necessary. For fun, if I had to guess when Tesla would have a version released that works 99% of the time, I’d say two years.