Former Apple executive Chris Lattner leaves Tesla after 6 months on the job
Former Swift lead Chris Lattner, who left Apple to head development of Tesla's Autopilot program earlier this year, is leaving the electric car maker after only six months of service.
Lattner unofficially announced his departure from the company in an impromptu post to Twitter on Tuesday.
"Turns out that Tesla isn't a good fit for me after all. I'm interested to hear about interesting roles for a seasoned engineering leader!" Lattner said. In a later post Lattner joked that he has seven years of experience with Swift and his resume is easy to find online.
Tesla later confirmed the departure in a statement to The Wall Street Journal, saying, "Chris just wasn't the right fit for Tesla, and we've decided to make a change."
As noted in the report, Tesla's Autopilot program has seen a number of high-profile departures over the past few months as CEO Elon Musk puts pressure on engineers to develop autonomous car systems. Musk's goal is to demonstrate a car that can drive itself by the end of 2017.
Currently, Tesla's Autopilot is semi-autonomous in nature and only assists drivers in performing important tasks like steering, braking and changing lanes. The system is not a replacement for human drivers, however, and drivers must monitor their car in a similar fashion to cruise control.
Lattner left Apple in January after playing a key role in the creation of the Swift programming language. It was quickly learned that Tesla had poached the software engineer to lead its Autopilot program.
In an interview conducted shortly after his departure from Apple, Lattner said he wanted to "accelerate the path to cars being appliances that solve people's problems."
Lattner unofficially announced his departure from the company in an impromptu post to Twitter on Tuesday.
"Turns out that Tesla isn't a good fit for me after all. I'm interested to hear about interesting roles for a seasoned engineering leader!" Lattner said. In a later post Lattner joked that he has seven years of experience with Swift and his resume is easy to find online.
Tesla later confirmed the departure in a statement to The Wall Street Journal, saying, "Chris just wasn't the right fit for Tesla, and we've decided to make a change."
As noted in the report, Tesla's Autopilot program has seen a number of high-profile departures over the past few months as CEO Elon Musk puts pressure on engineers to develop autonomous car systems. Musk's goal is to demonstrate a car that can drive itself by the end of 2017.
Currently, Tesla's Autopilot is semi-autonomous in nature and only assists drivers in performing important tasks like steering, braking and changing lanes. The system is not a replacement for human drivers, however, and drivers must monitor their car in a similar fashion to cruise control.
Lattner left Apple in January after playing a key role in the creation of the Swift programming language. It was quickly learned that Tesla had poached the software engineer to lead its Autopilot program.
In an interview conducted shortly after his departure from Apple, Lattner said he wanted to "accelerate the path to cars being appliances that solve people's problems."
Comments
I have a gut feeling that the Tesla software is a bowl of spaghetti in desperate need of a rewrite.
Not to worry, Tesla can license a Car OS from Apple soon. Tesla needs to get rid of that dated console UI anyway.
Then go back to Apple and build this...
The car of the future is already here. It's called a Smartphone. Think about it. If you were to clear the slate, look at the modern world and ask yourself, how would I design a transportation system given existing and soon-to-come technologies, like autonomous driving, real-time availability scheduling. Route optimization, etc, no way you'd conclude there should be a car, or two, in every garage. You'd create a technology/software infrastructure to allow individuals to call up the transportation they need (car, truck, van, etc) on-demand. And it would show up wherever they are, or wherever they are going to be, when it's needed. You'd be able to schedule transportation in advance, like the airport shuttles of yesteryear that you'd schedule a week in advance. Über pretty much killed that business, I expect.
Or schedule recurring transportation, such as to take the kids to soccer practice and back. In this case the transportation technology system might suggest a shared van service, that knows the schedules for local after school sports practice and offers up and constructs pick-up and drop-off routes based upon participation; a regular route to gather up the kids and deliver them. Accommodation for security will be considered when children are being transported without accompanying parents, such as real-time tracking and a constant open line of communication, both audio and video streaming from the vehicle to parent's smartphones.
The specific vehicle that arrives can be determined by number of passengers, whether you'll be transporting something large or just yourself, etc. The notion of owning, maintaining, accommodating parking requirements of, insuring, etc, a personal vehicle, for many people, has already begun to feel like 'the old paridigm.'
To create this infrastructure, you need route optimization software, that incorporates the real-time whereabouts of all vehicles in a local fleet. You need scheduling software. You need to deal with remaining charge/range of each vehicle out in service to know when a vehicle can accommodate an additional requested or scheduled route without running out of juice. You need to accommodate stand-by, where the vehicle drops someone off at a location and is requested to stand-by for an indeterminate time while the person goes into a store or bank to run an errand. In short, you need a very sophisticated set of interacting technologies to accommodate smooth operation of a transportation network that provides near immediate responsiveness to a population's constantly fluctuating needs.
If I were Tim Cook, this is exactly the way I'd envision the future, and this is what I'd set out to create. It's not so much about constructing vehicles yourself, but about getting sign-in from all vehicle manufacturers such that their vehicles can work within the envisioned transportation network. And that means that people who do own vehicles could lend them into their local autonomous transportation fleet in order to earn money (this has already been suggested by Musk and makes sense for a maker of vehicles to accommodate, as it helps him sell more Teslas direct to consumers). It means that new rental fleets will simply be staged in large metro areas, with one or more depots that the vehicles come back to for recharging, maintenance, cleaning, etc. And that means that there's a path forward for the rental companies, because they already have staging areas for their existing fleets. The big picture can be accommodated during a transition phase from the world we have today to a world where almost all transportation is shared and autonomous.
Extend this to trucking, inter-city bussing, etc, and the whole thing becomes a future that Apple could play a major role in developing. Without ever producing, on their own, a single vehicle.
Also key to this is that everything Apple needs to do to revolutionize transportation does not require Apple to do any work on autonomous driving, nor does Apple need to build a single vehicle model. Nope, Apple will want to own the end user interaction used to summon and schedule transportation, and it'll want to own the route optimization algorithms and server side scheduling and dispatch. And take a cut of every ride.
There will need to be some tech in each car to pick up the user interaction that began on a rider's smartphone or Watch, once the car arrives to pick up the rider. The car will need a voice interface to interact with the rider. The car will need to constantly ping its whereabouts to the dispatch and scheduling servers, along with its charge level, so that the dispatch system can determine its next pick up and determine when it needs to exit the active fleet and return to a nearby depot for recharging or maintenance. The car will need to contain sensors, like internal cameras, to monitor for left-behind packages, spilled coffee, etc, and report appropriately to riders or to dispatch. The car will need streaming audio/video capabilities to stream to parents when children are riding without adult accompaniment. All of this can be designed as a set of interfaces that automakers can implement in order to be compatible with Apple's dispatch and routing servers, and the vehicles might also be required to utilize Apple's mapping infrastructure.
Once verified as able to serve a ride request, the car is handed details on the location of the rider, and the rider's destination, and it can then utilize its own autonomous driving capabilities to serve the request. And all of this can integrate both driverless and human driven vehicles into the same service. So as vehicles are developed that are licensed for autonomous operation, these can be added to an existing Uber-like fleet of human driven vehicles, both serving together to form a centrally requested and directed/dispatched swarm serving a metrolitan area. Eventually, the human driven vehicles would all be replaced with autonomous vehicles, and the future will have arrived.
Agree, if Lattner really is interested in cars and autopilot, maybe Apple could use him on their project (whatever that turns out to be), haven't heard bad things about him from Apple or people at Apple so perhaps no hard feelings?
Musk is probably pushing too hard to get the AP 2.0 software done.
I'm not expecting carOS any time soon from Apple. Fortunately Cook brought Mansfield in last year to reign in TITAN. And now we finally get big hardware updates at WWDC, Now hopefully Apple can fix SIRI in the next six months before HomePod is released.
I think that it is more likely that the Tesla engineers did not like an outsider who is not familiar with the history of the code base coming to their rescue. Moreover, Tesla does not have time to rewrite their code from the ground up with artificial intelligence and machine learning at the core. They probably have their fast and furious ways of introducing new features requested by Elon Musk into their project and prefer piling on layers and layers of spaghetti onto it. I think they should have started a parallel project to modernize code.
I think a couple of facts are as follows:
1. Tesla's console UI is outdated and I suspect that the code behind the UI is outdated as well. For example, their UI has no usable voice recognition and the code behind the UI probably require a rewrite to be modernized with machine learning and artificial intelligence at the core and Tesla has no time for that now.
2. Chris has lead great open source work on LLVM, Clang, Swift etc... so he knows how to structure large scale software projects and work with others.
Since even Tesla thought he was a match, the reason for the fall out is probably culture, political and just plain push back against an outsider's view of the code base and future direction. You have to wonder why so many engineers tried this hot seat and failed.
Here is an interesting article: http://www.pcworld.com/article/260883/tesla_cto_talks_model_s_batteries_and_incar_linux.html
Time will tell.