Tesla updated its new autopilot hardware version to achieve "Full Self-Driving Capability"

author: Amanda   

Amanda Zhao

By Amanda Zhao

Recently, Tesla announced its new auto pilot feature on their official website. As it explains; "The system is designed to be able to conduct short and long distance trips with no action required by the person in the driver's seat. Your Tesla will figure out the optimal route, navigate urban streets (even without lane markings), manage complex intersections with traffic lights, stop signs and roundabouts, and handle densely packed freeways with cars moving at high speed. When you arrive at your destination, simply step out at the entrance and your car will enter park seek mode, automatically search for a spot and park itself. A tap on your phone summons it back to you." 

However, Tesla is also careful enough to state in bold lettering: "Please note that Self-Driving functionality is dependent upon extensive software validation and regulatory approval, which may vary widely by jurisdiction." The possibility to test the self-driving feature is expected to be released next year. 

By taking a look at the sensors mounted on the new self-driving feature, readers can grasp an overview from the following image:

Image from Tesla website

In terms of the hardware, all of those sensors can "see" a great deal. However, the real challenge regarding the software involves the question of whether the computer will be able to process the data acquired from all those sensors. Tesla Vision states it will implement an entirely new and powerful set of vision processing tools built into a deep neural network. 

A person's driving skills rely mostly on his or her driving experiences. Similarly, neural networks need to be smart enough to account for all of the decisions made to achieve the "Level 5" self-driving mode. 

Some would say that without many thousands of miles to demonstrate a car's reliability, they won't trust the autonomous system to give up control of the steering wheel. This notion is actually quite reasonable. That said, there are two major factors one might be led to believe a road test could help with. First, to recognize objects around the vehicle – meaning is there a deer standing in the middle of the highway or some foam or plastic? An autonomous driving system's software must be smart enough to know the difference. Second, risk analysis for when an extreme situation pops up out of nowhere. A good driver who is experienced will be better able to deal with extreme circumstances. (A National Highway Transportation Safety Administration study claimed that driver error is estimated as the cause of 94% of accidents). In an extreme situation, a human in control of the vehicle should choose a minimum damage path. 

Google recently announced their self-driving car has driven more than 2 million miles. Similarly, Tesla has a long way to go to accumulate enough experience for the Tesla Vision and the end results they wish to achieve. 

Amanda

Amanda Zhao is a Mechanical Engineer working at a high tech company in the Silicon Valley. She had her Master’s degree in University of Maryland with the focus of Medical robotics. With the curiosity of what’s happening around the ‘Centre of Universe’, she has the passion for writing tech news and sharing.