close
00c24cc6-4f03-11ea-9f9f-aaaa00151f43.jpg

Editor’s note: This article comes from the WeChat public account “Xinzhiyuan” (ID: AI_era), editor: Motoko, 36 氪 Authorized release Musk promises to have a complete “full-featured fully automatic driving” (FSD) function by the end of 2019Failed to recognize that it was at least “months” before the full FSD was launched.Musk said that he will organize a hackathon to solve the problem of autonomous driving, which shows that Tesla hopes to train neural networks to identify every object that may be encountered in the city and take corresponding actions to forcefully enter the city to drive.The promised fully-automated driving function failed to arrive as scheduled and Elon Musk told Tesla shareholders in February last year that the company would complete a fully-functional Full Self Driving (FSD) function by the end of 2019.According to The Verge, Musk told reporters in a telephone interview in February last year that there would be no problem with the release at the end of 19: “Now of course the driver is first required to supervise, because it will take billions of miles to reach theThe level of safety observed by the driver is needed again. Then we need to convince the regulators, which requires some steps … but as I said before, I am sure that we will release a fully automated driving package this year. “Full Self (Full SelfDriving (FSD) is the name of a Tesla software package that Musk claims will enable existing Tesla cars to drive from residential areas, highways, and city streets without any manual intervention.According to Business Insider, prior to last week’s fourth quarter earnings call, Musk acknowledged that Tesla would be at least “months” away from launching a “fully-functioning fully-autonomous drive.”In October 2019, Musk was forced to acknowledge that Tesla would miss his 2019 deadline and evaded his previous remarks.Musk recently announced that he will hold a “super fun AI party / hackathon” with Tesla engineers and some invitees.Musk will host an old-fashioned hackathon to solve the problem of autonomous driving.Taking a different approach, Musk wants to solve the problem with a hackathon. Let’s quickly review the recent conference call. In his original words: “Yes, it is fully functional. I mean, this car can be driven from home to work.It is likely that there is no need to intervene. Therefore, although it will still be supervised, it will be able to drive to the company. You enjoy high-speed autonomy on the highway and autonomy in medium-speed areas. For taxis, this actually means only at low speedThere are more traffic lights and parking signs. Therefore, the complete function means that it is likely to do this without manual intervention, but it will still be monitored. And I have said it several times, but it is often misunderstood asThere are three main levels of autonomy. Cars can drive autonomously, but sometimes require supervision and intervention. This is already fully functional. Perfect functionality does not mean that it can handle any situation anywhere on the planet (including every extreme situation). Functional perfection is onlyMeans most things. “According to Business Insider, Musk told the earnings callOr, the company’s goal now is to reduce ambitions (expectations): “Perfect functionality just means that there is a certain degree of autonomy that can be driven from home to the company without any intervention. But that doesn’t mean that all functions work well.”Musk and Tesla used this as a computer vision problem, quoting a Musk explanation in Teslarti’s latest report:” I think it may be months now. And, for autonomous driving and completeThe obvious thing about autonomous driving is that there is a lot of work to do on the basics of improving autonomy. “” As far as labeling is concerned, video is annotated in all eight cameras simultaneously. In terms of labeling efficiency, myMeaning, it can be said that there are three orders of magnitude improvement in labeling efficiency. For those who understand this, it is extremely important, so it has indeed made a lot of progress in this regard. “It sounds like Musk is talking aboutIt’s a tricky problem with computer vision: real-world annotation.However, we can vaguely feel that Tesla may train the neural network through violent methods so that it can identify all possible objects in the real world and take corresponding actions.This approach has potential flaws: because things don’t always seem so simple.Aside from Tesla, autonomous driving research requires careful consideration of the city’s infrastructure to better serve autonomous driving.At present, taxis in urban areas cannot be called unmanned ubers. It is not because regulators do not allow unmanned vehicles to hit the road, but because computer vision and lidar cannot completely solve the problem..

Tags : 36 氪 starterAutopilotBlack technologycapital marketcareducationEntrepreneurshipForefrontFrontlinegameGo publicInformationInternet of ThingsinvestmentNet red economyNew businessNew economyOccupationreal estatestockTechnologyTelematicsthe Internetunmanned