General Motors Acquires Lidar Specialist Strobe Inc.

3 weeks ago by Christopher Smith 28

Cruise Automation – Chevrolet Bolt EV

The automaker is one step closer to making self-driving cars a reality.

General Motors hopes to take bigger steps in the autonomous world, and to help make that happen the automaker has acquired California-based Strobe, Inc.

And no, this isn’t a company dedicated to strobe lights for Halloween parties, but it’s actually not that far off – Strobe specializes in LiDAR technology, something which is quickly becoming a key component of autonomous systems for automobiles. Strobe engineers will become part of GM’s Cruise Automation team.

“The successful deployment of self-driving vehicles will be highly dependent on the availability of LiDAR sensors,” said Julie Schoenfeld, Founder and CEO, Strobe, Inc. “Strobe’s deep engineering talent and technology backed by numerous patents will play a significant role in helping GM and Cruise bring these vehicles to market sooner than many think.”

Unlike radar systems that use radio waves to “see” an area, LiDAR uses laser pulses to generally accomplish the same task. The difference, however, is that LiDAR can create a far more detailed picture of its surroundings, which of course allows for more accurate processing of information. With a clearer picture of everything from cars to the road and pedestrians, LiDAR systems will play a pivotal role in the current and future development of autonomous cars.

That’s a segment GM hopes to lead. Last month, the Cruise Automation team unveiled what the automaker says is the world’s first autonomous car equipped with the necessary safety and redundancy systems required for operation without a driver. This alone isn’t ground breaking, but GM says the car is also able to be mass-produced, which brings us that much closer to driverless cars on the road.

“Strobe’s LIDAR technology will significantly improve the cost and capabilities of our vehicles so that we can more quickly accomplish our mission to deploy driverless vehicles at scale,” said Kyle Vogt, Founder and CEO, Cruise Automation.

Tags: , , ,

28 responses to "General Motors Acquires Lidar Specialist Strobe Inc."

  1. Chris O says:

    Guess that explains why GM’s director of autonomous vehicle integration Scott Miller recently attacked Elon Musk as being “full of crap” for claiming not to be needing Lidar. Somebody needed to justify this acquisition…

    1. SJC says:

      It is possible lidar is necessary and Musk is trying a confidence job.

      1. Vexar says:

        Yeah, I don’t know. I have at least 100k miles under my belt, and I’m not entirely sure what LIDAR is other than lasers and a lot of money. I’ve never needed it to get from point A to B in any car I’ve driven, and all I have is a couple eyeballs and ears as far as sensor equipment.

        Something tells me it boils down to processing the sensor data, not more data. Elon’s betting on the computing power and software, not the sensor variety.

        As much as he fears AI, and probably fairly so, the fact that so many of you out there think LIDAR is necessary tells me that AI has a long, long way to go before it can match what humans do, every day, and sometimes when they have to pee.

        1. Pushmi-Pullyu says:

          A self-driving car needs to build up a SLAM (Simultaneous Localization And Mapping) virtual reality, 3D map of its immediate environment. LiDAR is the easiest and fastest way to do that. LiDAR uses active scanning, “painting” the surroundings with a laser to provide reliable, instantaneous data on an object’s exact distance and shape.

          Neither radar nor optical recognition systems (using cameras) can provide this. Radar gives a much fuzzier image, and camera images require software to interpret what the camera “sees”. That takes a lot of processing power and time, and computer-driven optical object recognition has in practice been proven to be unreliable.

          Furthermore, cameras have the same limitations as the human eye at not being able to see in the dark; LiDAR and radar have no such limitations.

          What developers of autonomous driving systems should be aiming for isn’t to slavishly reproduce how a human being sees the road and drives on it. The goal should be developing the most capable, and safest, sensors and driving system. Frankly, I don’t see how that can be done without using LiDAR.

          Maybe Tesla will surprise me, but robotics specialists and software engineers have been trying for decades to improve optical image recognition, and they are still a long way away from it being reliable. The odds that Tesla can pull that off within a couple of years… well, I’m not very optimistic.

          1. Vexar says:

            So LIDAR is easier to deal with than image recognition. Yeah, I’ll agree with you there. Tesla is trying to do machine vision the hard way.

            1. Doggydogworld says:

              It’s not just reduced processing. Lidar gives a much higher degree of certainty. You really don’t want your self-driving system “guessing” if there’s a semi trailer in your path or not. It needs to know for sure.

          2. scottf200 says:

            So what do cars that have a dependency on LIDAR do in the rain or snow? Do they just pull over, do those taxis no longer run, do cars owners just get a message to drive manually because of weather. It is shocking how well the Tesla AP1 has worked in a large variety of rain storms … I would say it was seeing better than humans (me) in some cases. Of course, you hold on to the wheel even tighter and pay more attention but now you have radar to pick up slowing cars faster than humans (me) looking through the windshield (and especially at night|dusk)

            1. Doggydogworld says:

              Lidar doesn’t stop working in rain/snow/etc. The signal degrades. Camera images also degrade, especially at night. Even radar signal can degrades, though not nearly as much.

              Car with Lidar also have radar and camera. They still get more data than camera+radar alone, even in poor weather.

              When signals degrade self-driving cars will slow down, just as a human driver would (or should).

              1. Pushmi-Pullyu says:

                Exactly right, thanks.

                Yes, LiDAR return degrades somewhat in rain. But as Doggydogworld said, that doesn’t mean LiDAR stops working. Pretty much like you or me trying to drive while peering thru the rain.

                Will fully autonomous cars slow down when driving in heavy rain? Perhaps so. One thing is certain: Autonomous cars won’t ignore bad weather conditions and keep driving as if the roads were dry and clear, as all too many humans do!

                “The thing to keep in mind is that self-driving cars don’t have to be perfect to change the world. They just have to be better than human beings.” — Deepak Ahuja, CFO of Tesla Inc.

      2. wavelet says:

        It’s pretty clear neither Musk nor anyone else has religious convictions on this. Tesla is very much using software as a differentiator, and emphasizing partial autonomy ASAP; full Level-5 autonomy is unlikely to be legal outside test vehicles for 15-20 years at least.

        The various sensor types all have their (dis)advantages; LiDAR’s is price, and if a physically rotating sensor is used like the original Velodyne designs, potential long-term reliability.

        Pretty much every technical article about the sensor types I’ve read says the best results come from combining input from all available types.

        Tesla is trying to do as much as they can without LiDAR, for now; I’m sure they’ll integrate it when feasible, and that their SW takes that into account.

        1. Kdawg says:

          IIRC, Strobe’s design removes 95% of the cost of a LiDAR system.

    2. Pushmi-Pullyu says:

      Tesla is, I think, the only company which is spending significant resources to develop an autonomous vehicle while avoiding use of LiDAR.

      Whether or not Tesla can succeed at developing a system which is as safe as one which uses LiDAR for active scanning is, in my opinion, highly questionable.

      I don’t think there is any doubt that LiDAR would make development easier. It seems to me that Tesla is trying to cheap out by avoiding LiDAR. Solid state LiDAR systems are dropping price rapidly, so ultimately Tesla’s decision to avoid using LiDAR may appear short-sighted.

      1. Chris O says:

        Hmm, since you are not an autonomous driving engineer I think I will give Elon Musk’s opinion a lot more credit than yours. I can see the appeal of doing away with lidar which is expensive and seems to be implemented as an ugly rotating device on top of cars on test vehicles, one wonders how it could be mounted invisibly on production vehicles.

        1. theflew says:

          Elon’s main reason for not using Lidar was cost, but solid state Lidar from many vendors greatly reduces cost. One vendor is saying $100/sensor. So given autopilot is a $5k option that would give them some wiggle room on cost.

        2. Four Electrics says:

          I’ll take GMs word over Elan’s based on Cruise’s self driving progress in downtown San Francisco. Check out the videos; miles ahead of Tesla.

          P.S. Elon is not an engineer. His degrees are in physics and economics. He did not even found Tesla, and his hyperloop study was ghostwritten by actual engineers. His Model X decisions make that pretty obvious.

          1. Pushmi-Pullyu says:

            Elon Musk most certainly is a rocket engineer. He’s largely self-taught in that area of study.

    3. JeremyK says:

      I think the onus is on Musk, to prove out his NON-LIDAR based systems, not on those who chose to use them.

  2. Will says:

    Autopilot, oh no it’s GM

  3. HVACman says:

    I tie the relative paths GM and Tesla are taking to their relative financial conditions. GM’s strong positive cash and profit situation and deep engineering/production pool allows them to be selective about their Autonomy solutions. Their perception has been that outside vendors have developed the leading-edge (and patented) autonomous technology hardware/software solutions. And GM, with their cash, can afford to buy them in-whole and have exclusive access to it.Tesla – stock-valuation rich but cash poor – scraping for every $, robbing Peter to pay Paul, desperately scrambling to ramp up Model 3 production (and sales) before this revolving cash game goes bad – has to boot-strap their autonomous path, which means using what they can create in-house without a lot of cash. Guerrilla engineering. In Silicon Valley’s start-up economy, that may have an advantage, but at the global level, autonomy is a much bigger war, requiring hardware aircraft carriers, not software IED’s. Advantage GM.

    1. Doggydogworld says:

      Tesla can acquire companies using stock, so cash is no issue.

      1. HVACman says:

        Tesla can issue new stock and complete a stock-only-conversion buyout in lieu of cash only if:

        a) Current Tesla stockholders don’t mind even more stock value dilution.
        b) the stockholders of the company being purchased actually want diluted Tesla stock instead of cash.

        Until Tesla rights their financial ship and start to lives up to their current highly-inflated stock value, I don’t think we’ll see stock used as a carrot to entice companies to join the Tesla team.

        1. ModernMarvelFan says:

          They had no problems with TSLA bailing out SCTY with its stocks in a such “bad purchase” for TSLA. Why wouldn’t they approve a buyout with TSLA stock on something that is far more critical to TSLA’s future?

    2. Doggydogworld says:

      Also, Tesla can afford to throw more cash into R&D for something like this than GM, since the market doesn’t care about profit/loss at Tesla in the same way the do for GM.

      1. HVACman says:

        Tesla has to HAVE the cash before they can THROW the cash. That is their current problem. All throwable cash is flying into Model 3 production start-up burn pit. Until Model 3 sales and deliveries really take off and the cash squeeze backs off, they are SOL as far as new tech acquisitions. And they are missing some great opportunities.

  4. Get Real says:

    BS, if it wasn’t for EM’s aggressive push into self driving GM and virtually every other laggard OEM would still be slow walking their programs.

    Same is true of electrification.

    As far as GM being a global powerhouse, the fact that they are abandoning markets like Europe and right hand drive markets tells me that they are in fact shrinking globally while Tesla is expanding. Obviously GM is a much larger company then Tesla at this point but in the long-run–who knows.

    While LIDAR is promising it is severely degraded in rain, snow and fog. That alone means it will only contribute part of the time and in rainy/foggy/snowy climates it might be almost worthless.

    I do agree that EM is hardware agnostic.

    If SS Lidar comes down in price to be very affordable I could see Tesla adopting them in future builds at some point.

    1. HVACman says:

      RHD markets (Japan, GB, AU) and the EU are shrinking markets – aging population, stagnant economies, and population shrinkage. All the growing markets are LHD and most in mainland Asia. Watch where GM is growing and that is where the world auto market is growing.

      1. ModernMarvelFan says:

        Mostly right except for India and Countries such as Malaysia where population growth rate is far greater than China and much younger in age mix as well.

  5. Get Real says:

    Ridiculous, the EU market alone is bigger then the US market.

Leave a Reply