Tesla CEO Elon Musk made another rumor, claiming that his electric car will have auto-driving capabilities by the end of 2017, driving all the way from Los Angeles to New York. The Seeking Alpha website believes that Tesla has the confidence to develop an autonomous driving system in a short period of time thanks to Nvidia.
Seeking Alpha reported on the 24th that Tesla originally cooperated with Israeli mobile phone Mobileye to use EyeQ3 chip. Its electric vehicle has automatic emergency braking, collision warning, and maintenance of lanes. However, Tesla and Mobileye broke up and changed to Nvidia system. The original advanced driver assistance system (ADAS) is completely ruined.
In this case, why does Tesla dare to say that there will be automatic driving at the end of next year? According to the article, Nvidia has developed a revolutionary self-driving system that collects human-real driving movies and uses steering wheel steering angles to train a neural network (NN); that is, every driving scene, Have the correct answer, let NN learn to play the steering wheel angle; when paired with the departure lane, how to correct the information of the return line, let the computer master the driving skills.
The hallmark of this approach is that NN can “master†the driving skills. Prior to this, the industry mostly subdivided the self-driving into multiple projects, such as identifying objects, roads, etc., and then combining them, unable to confirm whether the computer really caught the driving car. Nvidia's new method is simple and fast, with excellent results in less than a year. The car can be driven automatically under a variety of conditions. If it is based on this system, it should be able to quickly increase the self-driving ability. The article speculates that Tesla is in this phase, and then abandoned the old love Mobileye and Nvidia cooperation.
The article alleges that Tesla uses Nvidia technology to develop a self-driving model that can learn from human actual driving conditions and reactions, and accelerate research and development. However, the article warns that the problem is that machine learning does not know when it will run into a bottleneck, thinking that NN can learn to drive smoothly and is too optimistic. The threshold for self-driving is extremely high, and the NN learning process may be stalled and difficult to reach. The article pointed out that at present, Google's self-driving driving technology is still not as good as humans, and Tesla is far behind.
We accept Personalized logo Wireless Charger Pad Service. Let our team of experts design and build your very own Wireless charger Pad.All you need to do is send us a photo or sketch, we will then send you a FREE mockup of your design for approval.We are a professional Chinese manufacturer of Wireless charger Pad , and look forward to your cooperation!
China Wireless Charger Pad,Qi Wireless Charger Pad,Fast Wireless Charger Pad,Fast Qi Wireless Charger Pad Manufacturer
Reteck Electronic Co., Ltd. , https://www.reteck.com