Musk driverless cars may seem crazy, actually very reliable-IT information
Musk is the CEO of Tesla and SpaceX, he had made bold predictions for technology development, and outlines deadlines, time seems impractical. Once, Thomas g on a promise to let everyone use electric vehicles to put humans on Mars, in the pursuit of these crazy goals, Thomas g is known for, only time in sight.
Now, musk promises again in 2017, will before the end of production of Tesla's cars can automatically, no need to drive, the car can be opened from Los Angeles to New York. Many large enterprises are developing driverless cars, musk is given time beyond the opponent. Ford plans to launch in 2021 driverless cars, Baidu is set in the year 2019. Google did not give a specific time, is most likely the year 2021. GM also set in this year. Please note that automatic driving this car only in a limited area, is not a passage to China.
Musk can really achieve the aim of dream? Let our in-depth analysis:
Declaration
See Tesla's website will find that users can choose when buying a car, "automatic driving function", Tesla declared: "all you have to do is got into their car and tells the car where you want to go on the trip. "Actually the car not to, at least not yet. Mr MASKEY has not introduced the necessary software, early Tesla car first Autopilot don't even support. On the highway, Autopiolot can keep Tesla cars in the driveway, keep a safe distance with other cars.
Tesla has released a video associated with Model x SUV, it installed a new prototype of the hardware and software, you can automatically drive in Palo Alto. Looks amazing, in fact, even the concept has not yet been verified.
Unmanned technology expert at Carnegie Mellon University Laji·lakuma (Raj Rajkumar): "cars in the San Francisco area, do not edit the video, this may make more sense, because we see how cars have the ability. ”
If Tesla cars really as powerful as said in the video, it can really own accessible throughout the United States? Driverless car expert at the University of Southern California members of the IEEE, Geoffrey Miller (Jeffrey Miller) said: "if developed for a further year, may see similar things happen. ”
Looking back on the past record will find musk all missed time. Roadster and Model s, Model x, Model 3, Autopilot, without a product not to postpone, Tesla driverless cars could also be delayed.
Conclusion: Thomas g if the promise, of course, is an encouraging thing, but from the past record, how strong his sense of time is not.
Data
To develop a car, so that it can deal with any situation, need to teach car lot. In this regard, Google in the lead position, it already has a 60-car convoy of driverless cars, drove more than 2 million miles, most of the mileage in the city. Tesla claimed to have one advantage: its car in Autopilot mode 222 million miles, around the world, collecting large amounts of data.
Miller said: "Tesla is access to large amounts of data, mainly because it sells a lot of cars. "The problem is that Tesla has collected what data is uncertain. Google uses cameras, radar and laser sensors to collect data, alongside a human engineer supervision.
Tesla Motors is always online, don't assume that all data are useful. United States Princeton University's former computer engineer, driverless cars software startup founder xiaojianxiong (transliteration: Jianxiong Xiao) says: "in a car in the course of many data may be discarded. ”
Conclusion: Tesla have more data, more diverse, these data may be enough to help the company go beyond the competition.
Technology
Many car manufacturers to develop driverless car uses laser radar sensors, laser radar, laser pulse, drawing precise images to the surrounding environment. Musk against the technology, because the price is too high. Google engineers have said they installed laser radar equipment cost about $ 80,000.
Tesla Motors equipped with 8 cameras, a forward radar, musk that such devices provide perspective as well.
Miller said: "laser radar is a great technology, but it includes mobile components, sometimes it may fail, and the price is very high. "On rainy days, snowy or foggy weather, radar is not very practical. No Tesla laser radar can complete the mission? Tesla said: absolutely.
Not everyone thinks so. Xiaojianxiong said: "did Tesla really to find solutions? I doubt it, if it doesn't have a laser radar, it is difficult to ensure security. "He believes that Tesla could develop a car, it is human driven cars safer, but make sure that the 100% security select laser radar.
To convert sensor data into driving instruction, installed Nvidia Tesla Motors Drive PX 2 super computer, using advanced learning technology, cars can learn on your own. Car camera, Tesla developed software to control, before the software is provided by Mobileye, Mobileye has abandoned the Tesla, it believed that Tesla was too radical.
Conclusion: Tesla didn't use lidar, this unconventional, but might work.
The law
In Tesla's cars reservation page contains a warning message: "Please note that drone function is based on a broad definition of software validation and regulatory approval, because the judicial system there may be great differences. ”
In the United States, federal and State regulators to slow, there is no effective supervision of unmanned technologies, in other places, no agency has prepared Tesla can be submitted to the certification of Level 5 system.
Tesla's escalating the drone of cars feature 1.1 upgrade. In an upgrade on a car may improve the ability to identify traffic signals, another upgrade, may allow empty cars looking for parking spaces. The problem is that when the vehicle driverless cars, how existing and future legal regulations?
Conclusion: the only legislative approval before the robot car on the open road, musk is not willing to wait, but he may just have to wait.
Ambition
Each year, 1.2 million people were killed in a car accident, United States exceeds 30,000. Mr MASKEY said: "because of this, we should put them fully autonomous driving vehicles, from a security perspective, we believe that this car 1 time times safer than cars, could be even higher. ”
Some people expressed concern that too aggressively might impede progress. Laji·lakuma, a professor at Carnegie Mellon University, said: "Tesla is too reckless, desperate for immature technology to market, this technology is not ready for deployment, Tesla may be a cause for concern, because the technology could lead to crashes, resulting in death, social conflict, can get even tighter regulation. ”
Autopilot system there is a problem, but there are no serious consequences. Assistant Bryant Walker Smith, a law professor at the University of Southern California, said: "the increasing serious car accident, but it did not appear to be changing. "Germany Tesla regulators urged not to use the" Autopilot "terminology, because the wording there is exaggerated. Tesla is opposed.
Tesla's cars have the autopilot functions, these cars have been sold, although the functionality is not as strong as Mr MASKEY said. Now the market will prove to us, what do you need this kind of technology. Tesla's early users seem to like, Model 3 target market is the key, it is targeted to public users. Public is willing to believe that computer? Willing to believe that musk? Or let the sales figures speak for themselves.
Conclusion: the musk's ambition has not been achieved, but he also did not fail.
马斯克的无人驾驶汽车看起来很疯狂，其实很靠谱 - IT资讯
马斯克是特斯拉和SpaceX的CEO，他曾经对技术的发展进行过大胆的预测，而且还划定了截止期限，时间看起来有些不切实际。曾经，马斯克承诺要让每一个人用上电动汽车，要让人类登上火星，为了追求这些疯狂的目标，马斯克闻名天下，只是实现的时间遥遥无期。
现在，马斯克再次承诺，在2017年年底之前将会生产可以自动行驶的特斯拉汽车，不需要人驾驶，汽车可以自己从洛杉矶开到纽约。许多大企业都在开发无人驾驶汽车，马斯克给出的时间超越了对手。福特计划在2021年推出无人驾驶汽车，百度定在2019年。谷歌没有给出一个明确的时间，极可能是2021年。通用也定在这一年。请注意，这种汽车只能在有限的区域自动行驶，并非通行中国。
马斯克真的能够实现远大的梦想吗？让我们深入分析一下：
宣言
查看特斯拉官网就会发现，购买汽车时用户可以选择“全自动驾驶功能”，特斯拉宣称：“你所要做的就是钻进汽车，告诉汽车你要去哪里就行了。”事实上汽车做不到，至少目前还做不到。马斯克还没有推出必要的软件，早期的特斯拉汽车连第一代Autopilot都不支持。在高速公路上，Autopiolot可以让特斯拉汽车呆在车道上，与其它汽车保持安全距离。
特斯拉曾经公布一段与Model X SUV有关的视频，它安装了新的硬件和原型软件，可以在Palo Alto自动行驶。看起来很惊人，实际上连概念都还没有验证过。
卡耐基梅隆大学无人驾驶技术研究专家拉吉·拉库马（Raj Rajkumar）认为：“让汽车自己在旧金山市区行驶，不要对视频进行编辑，这样可能更有意义，因为我们可以看到汽车到底具备怎样的能力。”
假设特斯拉汽车真的像视频所说的那样强大，它真的能够自己通行全美国吗？IEEE成员、南加州大学无人驾驶汽车专家杰弗里·米勒（Jeffrey Miller）认为：“如果再开发一年，也许会看到类似的事情发生。”
回看过去的记录，就会发现马斯克老是错过时间。Roadster、Model S、Model X、Model 3、Autopilot，没有一款产品不推迟的，特斯拉无人驾驶汽车也可能会推迟。
结论：马斯克如果真的兑现承诺，自然是一件鼓舞人心的事，但是从过往记录看，他的时间观念不怎么强。
数据
要开发一辆汽车，让它可以应付任何情况，需要教会汽车许多东西。在这方面，谷歌处在领先位置，它已经拥有一个由60辆无人驾驶汽车组成的车队，行驶了200多万英里，大多的行驶里程是在城市进行的。特斯拉声称自己有一个优势：它的汽车在Autopilot模式下行驶了2.22亿英里，遍及全世界，收集了大量的数据。
米勒称：“特斯拉之所以获得大量数据，主要是因为它销售了许多汽车。”问题在于特斯拉到底收集了什么数据并不明朗。谷歌用摄像头、雷达、激光传感器收集数据，旁边还有人类工程师监督。
特斯拉汽车是一直在线的，不要认为所有的数据都有用。美国普林斯顿大学前计算机工程师、无人驾驶汽车软件创业公司创始人肖健熊（音译：Jianxiong Xiao）认为：“在汽车行驶的过程中，许多数据可能会丢弃。”
结论：特斯拉拥有的数据更多，更加多样化，这些数据也许足以帮助公司超越竞争对手。
技术
许多汽车制造商开发无人驾驶汽车时使用的是激光雷达传感器，激光雷达会发射激光脉冲信号，给周边环境绘制精准图像。马斯克反对这一技术，因为价格太高。谷歌工程师曾经透露说，他们安装的激光雷达设备造价约为8万美元。
特斯拉汽车安装了8个摄像头、一个向前的雷达，马斯克认为这种设备提供的视野同样出色。
米勒认为：“激光雷达是一种很棒的技术，不过它包括了移动组件，有时可能会失效，而且价格也很高。”在雨天、雪天或者有雾的天气，雷达不太实用。没有激光雷达特斯拉能够完成使命吗？特斯拉认为：绝对可以。
不是所有人都这样认为。肖健熊说：“特斯拉真的能够找到解决方案吗？我相当怀疑，如果没有激光雷达，很难保证安全。”他认为，特斯拉可能会开发一款汽车，它比人类驾驶的汽车更安全，但是要确保100%安全只有选择激光雷达。
为了将传感器数据转化为驾驶指令，特斯拉汽车安装了Nvidia Drive PX 2超级电脑，利用深度学习技术，汽车可以自己学习。至于汽车的摄像头，特斯拉用自己开发的软件控制，之前的软件由Mobileye提供，但是Mobileye抛弃了特斯拉，它认为特斯拉太激进了。
结论：特斯拉没有使用激光雷达，这种做法不合常规，但是可能行得通。
法律
在特斯拉的汽车预定网页上有一则警告消息：“请注意，无人驾驶功能是根据广泛的软件确认和监管核准界定的，因为司法系统不同可能会存在很大的差异。”
在美国，联邦和州监管机构步伐缓慢，还没有对无人驾驶技术进行有效监管，在其它地方，也没有什么机构已经做好准备，可以对特斯拉提交的Level 5系统进行认证。
特斯拉不断升级汽车的无人驾驶功能，一点一点升级。在某一次升级中，汽车可能会提高识别交通信号的能力，在另一次升级中，可能会允许空车寻找停车位置。问题在于，当汽车变成无人驾驶汽车，现有以及未来的法律如何监管？
结论：只有法律批准了，才让机器人汽车开上公路，马斯克不愿意等待，但是他也许只能等待。
野心
每年有120万人因为车祸丧生，美国超过3万。马斯克说：“正因如此，我们应该将汽车变成完全自主行驶的车辆，从安全层面看，我们相信这种汽车比人类驾驶的汽车安全一倍，可能更高。”
一些人表示担忧，太过激进可能会阻碍技术的前进。卡内基梅隆大学教授拉吉·拉库马表示：“特斯拉太鲁莽，急于将不成熟的技术推向市场，这种技术还没有为部署做好准备，特斯拉的做法可能会引起人们的担忧，因为技术可能会导致撞车，导致死亡，让社会抵触，甚至还可能招来更严格的监管。”
Autopilot系统虽然存在问题，但是没有酿成严重的后果。南加州大学助理法学教授Bryant Walker Smith说：“严重的车祸不断增加，但是事情似乎并无变化。”德国监管机构要求特斯拉不要使用“Autopilot”术语，因为这一措辞存在夸大之嫌。特斯拉持反对态度。
特斯拉的汽车具备自动驾驶功能，这些汽车已经销售，虽然功能没有马斯克所说的那样强大。现在市场会向我们证明，到底人们需不需要这样的技术。特斯拉的早期用户似乎很喜欢，Model 3所瞄准的市场才是关键，它所针对的是大众用户。公众愿意相信电脑吗？愿意相信马斯克吗？还是让销售数据来说话吧。
结论：马斯克的野心还没有实现，但是他也没有失败。