Tesla Developing Proprietary Chip For Autonomous Driving Use

 

When Tesla CEO Elon Musk gets an idea into his head, he is like a dog with a bone. “Single minded” does not begin to describe his determination to get what he wants. Sometimes, that narrow focus can be draining for those working with him.

Tesla Autopilot

Musk made a promise earlier this year that Tesla automobiles would be fully autonomous within two years. Shortly thereafter, Chris Lattner, a former superstar tech guru at Apple, severed his relationship with the company. Lattner had been in charge of the autonomous driving program since Sterling Anderson left the company at the end of 2016.

In a statement, Lattner said, “In the end, Elon and I agreed that he and I did not work well together and that I should leave, so I did.” He went on to say that the Autopilot team was involved in a stressful transition period and was struggling with attrition. The company began equipping all its cars with an all new collection of radar, camera, and ultrasound sensors — a package known as Hardware 2 — in October of last year.

Musk promised customers the HW 2 equipped cars would have all the self driving capabilities of older cars with the Hardware 1 package by the end of 2016. In fact, it took a lot longer than that. It was not until Consumer Reports downgraded the Model S because its emergency forward braking system was not yet fully operational that Tesla finally made good on its promise.

Now reports have surfaced that Tesla is working on designing its own proprietary computer chip for use in its autonomous driving systems. At present, it uses an advanced supercomputer from Nvidia to provide the massive amount of computing power required to make the millions of calculations per second necessary.

But Tesla is wary of outside suppliers and likes to do as many operations as possible in house, where it can control both performance, timeliness, and price itself. And the Nvidia unit is not specifically designed solely for autonomous driving duties. The chip Tesla is said to be working on what would narrowly targeted solely for such use.

The reports suggest that Tesla is collaborating with AMD on development of the new chip, but Sanjay Jha, CEO of AMD will only confirm that his company is working with several automakers on dedicated computer chips.

The AMD connection is buttressed by the fact that Jim Keller, who previously worked for AMD, has now taken over from Chris Lattner as head of the Autopilot unit at Tesla. Tesla has also hired several other AMD engineers to work with Keller, including director Ganesh Venkataramanan, principal hardware engineer Bill McGee and system circuit design lead Dan Bailey.

Elon has proclaimed that a Tesla Model  S will leave Los Angeles, drive itself to New York City, and park, all without intervention from a human driver before the end of 2017. That suggests the 50 people assigned to the Autopilot program probably aren’t going to be spending much time with their families for a while.

Adding a layer of intrigue to the story is another report that Tesla has developed yet another collection of hardware known internally as Hardware 2.5. This goes against Musk’s claim last year that cars with HW 2 would have all the sensors needed to permit full autonomous operation once regulations and software caught up with the system’s potential.

When asked about the new hardware configuration by The Verge, a Tesla spokesperson said, “The internal name HW 2.5 is an overstatement, and instead it should be called something more like HW 2.1. This hardware set has some added computing and wiring redundancy, which very slightly improves reliability, but it does not have an additional Pascal GPU.”

The person went on to say that cars with the existing HW 2 package would be upgrade at no cost to the customer if the new configuration proved more capable in real world testing. One of the features of Tesla is that is continues to improve its cars continuously and incorporates those improvements into production cars as soon as possible, rather than waiting for annual model changes the way traditional car companies do.

Elon Musk continues to overpromise and underdeliver. In theory, that is just the opposite of how a successful company is built. But he always gets there eventually and those who maintain their faith in him are amply rewarded with some of the safest, most innovative cars available.

In the grand scheme of things, whether a Tesla drives itself cross country this year or next won’t be significant. What will be significant is that a Tesla will have done it first. Every one knows the name of the first person to walk on the moon. Few know the name of the second person to do it. Elon Musk definitely knows the importance of being firstest with the mostest.

Source: CNBC





About the Author

I have been a car nut since the days when Rob Walker and Henry N. Manney, III graced the pages of Road & Track. Today, I use my trusty Miata for TSD rallies and occasional track days at Lime Rock and Watkins Glen. If it moves on wheels, I'm interested in it. Please follow me on Google + and Twitter.
  • James Rowland

    Global Foundries has denied this report and AMD declined to comment, so this should probably be filed under “rumours” for the moment.