Apple AI director talks advances in machine learning & mapping for self-driving car platfo...

Posted:
in Future Apple Hardware
Apple's director of AI research, Ruslan Salakhutdinov, gave peers a small glimpse into the company's self-driving platform this week, discussing some internal projects at the NIPS machine learning conference.




While one of the projects -- LiDAR object detection -- was detailed in a November research paper, Salakhutdinov also went into previously unpublicized areas, according to Wired. The company's camera-based recognition system, for instance, can discern objects even when lenses are obscured by rain, and identify pedestrians on the side of the road even when they're partially hidden by parked vehicles.

"If you asked me five years ago, I would be very skeptical of saying, 'Yes you could do that,'" Salakhutdinov commented.

The director also talked up Apple's work on dynamic decision making by cars -- such as how to avoid a pedestrian -- and its use of "SLAM," simultaneous localization and mapping, a technology some autonomous machines employ to maintain a sense of direction.

Apple was further said to be creating 3D maps of cities, including details like traffic lights and road markings. Some of this data is presumably being collected by Apple's autonomous test vehicles, but still more could be coming from the Apple Maps vehicles touring cities around the world, which the company has yet to fully explain.

The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services. Before then it should begin running its internal "PAIL" (Palo Alto to Infinite Loop) shuttle.

Comments

  • Reply 1 of 13
    SoliSoli Posts: 8,403member
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    repressthis
  • Reply 2 of 13
    Rayz2016Rayz2016 Posts: 4,442member
    Soli said:
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    No technical reason why it shouldn’t, but Apple is about skating to where the puck is, and I’m not sure they see car ownership as the future. 

    http://fortune.com/2016/03/13/cars-parked-95-percent-of-time/


    repressthisGeorgeBMac
  • Reply 3 of 13
    Rayz2016 said:
    Soli said:
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    No technical reason why it shouldn’t, but Apple is about skating to where the puck is, and I’m not sure they see car ownership as the future. 

    http://fortune.com/2016/03/13/cars-parked-95-percent-of-time/


    OK, so remain conceptual as long as you can (= the easiest escape from market inquiries on actual product or service)
    edited December 2017
  • Reply 4 of 13
    Rayz2016Rayz2016 Posts: 4,442member
    Soli said:
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    I forgot to ask: what do you see as the ultimate goal?
  • Reply 5 of 13
    SoliSoli Posts: 8,403member
    Rayz2016 said:
    Soli said:
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    I forgot to ask: what do you see as the ultimate goal?
    I'd think that their system would work for "all other kinds of driving and vehicle ownership, and machines that aren't road-based," so I'd expect their ultimate goal to at least include that level of pervasiveness.
  • Reply 6 of 13
    Rayz2016 said:
    Soli said:
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    No technical reason why it shouldn’t, but Apple is about skating to where the puck is, and I’m not sure they see car ownership as the future. 

    http://fortune.com/2016/03/13/cars-parked-95-percent-of-time/


    Apple's DNA isn't in rental services though. They have always been about personal items and people taking ownership of what they sell.
    Ride sharing and people not owning cars doesn't seem like it fits unless they made the whole car cabin a platform and the service was all the plugin and consumable parts that moved cabins from point A to B.  So cabin could be self driving car, turn up to a hyperloop get loaded and be in another city for holidays with their own car and only packing once.
  • Reply 7 of 13
    Rayz2016Rayz2016 Posts: 4,442member
    Soli said:
    Rayz2016 said:
    Soli said:
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    I forgot to ask: what do you see as the ultimate goal?
    I'd think that their system would work for "all other kinds of driving and vehicle ownership, and machines that aren't road-based," so I'd expect their ultimate goal to at least include that level of pervasiveness.
    Oh, I see. That’s fair enough. I just don’t see them building their own car to sell to people. 
  • Reply 8 of 13
    Rayz2016Rayz2016 Posts: 4,442member

    mattinoz said:
    Rayz2016 said:
    Soli said:
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    No technical reason why it shouldn’t, but Apple is about skating to where the puck is, and I’m not sure they see car ownership as the future. 

    http://fortune.com/2016/03/13/cars-parked-95-percent-of-time/


    Apple's DNA isn't in rental services though. They have always been about personal items and people taking ownership of what they sell.
    Ride sharing and people not owning cars doesn't seem like it fits unless they made the whole car cabin a platform and the service was all the plugin and consumable parts that moved cabins from point A to B.  So cabin could be self driving car, turn up to a hyperloop get loaded and be in another city for holidays with their own car and only packing once.
    Apple’s DNA is a seamless user experience. They have been doing rental services for years, but because it is seamlessly wedded to their platform you may not have realised it. 

    Apple Music is a rental service. 
    iCloud storage is a rental service. 
    iTunes Match is a rental service. 

    You could even even argue that their iPhone upgrade program is a rental service. 

  • Reply 9 of 13
    SoliSoli Posts: 8,403member
    Rayz2016 said:
    Soli said:
    Rayz2016 said:
    Soli said:
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    I forgot to ask: what do you see as the ultimate goal?
    I'd think that their system would work for "all other kinds of driving and vehicle ownership, and machines that aren't road-based," so I'd expect their ultimate goal to at least include that level of pervasiveness.
    Oh, I see. That’s fair enough. I just don’t see them building their own car to sell to people. 
    My statement didn't include any implication that they would or wouldn't build their own cars. Whose cars are Didi using right now? How would Apple's system interact with Didi? Either it's an aftermarket solution for a variety of cars, partnering with actual automobile makers to use Apple's system, or Apple creating their own vehicles that use Apple's system. In any of those scenarios limiting its use to just Didi is not limiting the potential market since Didi uses cars that are just like all the other cars sold and used by everyone else. 

    BTW, Apple building their own car is more inline with their history (but I also don't see that happened), but what's not inline with their history is a built-in or aftermarket system for other vendors to use, at least not with HW. I think no matter what they do it'll be unique from all their CE efforts.
  • Reply 10 of 13
    Rayz2016Rayz2016 Posts: 4,442member
    Soli said:
    Rayz2016 said:
    Soli said:
    Rayz2016 said:
    Soli said:
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    I forgot to ask: what do you see as the ultimate goal?
    I'd think that their system would work for "all other kinds of driving and vehicle ownership, and machines that aren't road-based," so I'd expect their ultimate goal to at least include that level of pervasiveness.
    Oh, I see. That’s fair enough. I just don’t see them building their own car to sell to people. 
    My statement didn't include any implication that they would or wouldn't build their own cars. Whose cars are Didi using right now? How would Apple's system interact with Didi? Either it's an aftermarket solution for a variety of cars, partnering with actual automobile makers to use Apple's system, or Apple creating their own vehicles that use Apple's system. In any of those scenarios limiting its use to just Didi is not limiting the potential market since Didi uses cars that are just like all the other cars sold and used by everyone else. 

    BTW, Apple building their own car is more inline with their history (but I also don't see that happened), but what's not inline with their history is a built-in or aftermarket system for other vendors to use, at least not with HW. I think no matter what they do it'll be unique from all their CE efforts.
    I think that’s where we disagree. 

    They’ll do anything else, I just don’t see them building their own car to sell to people because I don’t think Apple sees individual car sales as a long term prospect. 

    But sellng the system to other car manufacturers? Yup, why not? Using it to manage a rental fleet (their own or others)? Hell, yes. 

    And if Apple is about anything then it’s about change. Why do you think their “fans” scream so much? Apple’s history is about extending its ecosystem as far as possible. 

    What I don’t know is how this would work. I imagine that all AI car systems will be using similar hardware sensor kits, so does Apple need to supply the hardware?
  • Reply 11 of 13
    I guess if Apple’s goal is to partner with Chinese or Indian electric vehicle manufacturers that would enable them to focus on software and not get entangled with the vehicle end of things.
  • Reply 12 of 13
    SoliSoli Posts: 8,403member
    Rayz2016 said:
    Soli said:
    Rayz2016 said:
    Soli said:
    Rayz2016 said:
    Soli said:
    The ultimate goal of Apple's efforts is unknown, but may involve a platform for ride-hailing services.
    I keep reading that, but why? If the platform can work for ride-sharing, why can't it work for all other kinds of driving and vehicle ownership, and machines that aren't road-based? As an initial goal, that seems like a great start, which explains their ties to Didi, but I don't see that has an ultimate goal.
    I forgot to ask: what do you see as the ultimate goal?
    I'd think that their system would work for "all other kinds of driving and vehicle ownership, and machines that aren't road-based," so I'd expect their ultimate goal to at least include that level of pervasiveness.
    Oh, I see. That’s fair enough. I just don’t see them building their own car to sell to people. 
    My statement didn't include any implication that they would or wouldn't build their own cars. Whose cars are Didi using right now? How would Apple's system interact with Didi? Either it's an aftermarket solution for a variety of cars, partnering with actual automobile makers to use Apple's system, or Apple creating their own vehicles that use Apple's system. In any of those scenarios limiting its use to just Didi is not limiting the potential market since Didi uses cars that are just like all the other cars sold and used by everyone else. 

    BTW, Apple building their own car is more inline with their history (but I also don't see that happened), but what's not inline with their history is a built-in or aftermarket system for other vendors to use, at least not with HW. I think no matter what they do it'll be unique from all their CE efforts.
    I think that’s where we disagree. 

    They’ll do anything else, I just don’t see them building their own car to sell to people because I don’t think Apple sees individual car sales as a long term prospect. 

    But sellng the system to other car manufacturers? Yup, why not? Using it to manage a rental fleet (their own or others)? Hell, yes. 

    And if Apple is about anything then it’s about change. Why do you think their “fans” scream so much? Apple’s history is about extending its ecosystem as far as possible. 

    What I don’t know is how this would work. I imagine that all AI car systems will be using similar hardware sensor kits, so does Apple need to supply the hardware?
    I'm not sure where we disagree except for your statements about a very limited market segment that is most of the market on the table. Again, I made no declaration that would require Apple to build their own cars or not build their own cars to get the same market options.

    To put another way, lets say Didi using a car from, say, Ford. You've stated that their Didi vehicle will use Apple's tech, which means an aftermarket solution or Apple also partnering with Ford to get this built-in. So why only do this through Didi as the ultimate goal? Why not allow anyone that wants to have Apple's tech get it? Your link to cars sitting idle 95% of the time is irrelevant since Apple isn't selling fuel. A sale is a sale. It would be like Apple limiting Apple Music sales to only those that listen to 20+ hours a week. It simply wouldn't be good business to deny everyone else, including those that want to listen for less than one hour a week.
  • Reply 13 of 13
    Detailed maps of fixed objects could be used to drive a system of electronic BEACONS that broadcast live information to cars about pedestrian movement, road and traffic conditions ahead, alternative routes, commercial ads for nearby retailers, etc. Apple BEACONS could also provide clear visual coverage of the route, during darkness or bad weather. If the government is not going to create an electronic backbone for self-piloting cars, then someone will.
Sign In or Register to comment.