Wednesday, August 21, 2019

Automotive News: Porsche-Trieye, Koito-Daihatsu

Globes: Porsche has invested in Israeli SWIR sensor startup Trieye. “We see great potential in this sensor technology that paves the way for the next generation of driver assistance systems and autonomous driving functions. SWIR can be a key element: it offers enhanced safety at a competitive price,” says Michael Steiner, Member of the Executive Board for R&D at Porsche AG.

Porsche $1.5M investment is a part of Series A round extension from $17M to $19M. TriEye's SWIR technology is CMOS-based, said to enable the scalable mass-production of SWIR sensors and reducing the cost by a factor of 1,000 compared to current InGaAs-based technology. As a result, the company can produce an affordable HD SWIR camera in a compact format, facilitating easy in-vehicle mounting behind the car’s windshield.


Nikkei: Daihatsu low cost car Tanto released it July 9, 2019 features adaptive headlight technology from Koito. This is the first appearance of such a technology in low cost car, probably signalling a beginning of the broad market adoption:

"When a light-distribution-changeable headlight detects an oncoming or preceding vehicle at the time of using high beam, it blocks part of light so that the light is not directed at the area in which the vehicle exists.

In general, it uses multiple horizontally-arranged LED chips for high beam and controls light distribution by turning off LED light that is directed at an area to which light should not be applied.

With a stereo camera set up in the location of the rear-view mirror, it recognizes oncoming and preceding vehicles. By recognizing the color and intensity of light captured by the camera, it judges whether the source of the light is a headlight or taillight.

When it recognizes a vehicle, LEDs irradiating the area of the vehicle are turned off. It can recognize a headlight about 500m (approx 0.31 miles) away.
"

13 comments:

  1. The "SmartBeam" technology was pioneered by Photobit and Gentex in the late 1990's and progressed to being standard technology in many upper end vehicles in the past 10-15 years (the automotive market is very slow). Glad to see it migrating down to less expensive cars. It is a great safety feature, which was Fred Bauer's intention (CEO, Gentex).

    ReplyDelete
  2. This is amazing - can we make this a legal requirement like tomorrow?

    ReplyDelete
  3. there is a surprising ammount of news around near/shortwave infrared detection with "alternative" technology other than ingaas recently. in this field there are restrictions due to potential military use, isn't it? does this limit the possible properties like resolution if the sensor is supposed to be sold without military export restriction?

    ReplyDelete
  4. @anon, we can also flip the argument and ask - if we keep improving Si SPADs PDE at 905nm for LiDARs, they may also become sensitive in SWIR like at 1064nm, etc. in 5-10years, and then they may also come under export restrictions? So, we should stop to improve SPADs before ITAR guys take a notice.

    What do you think?

    ReplyDelete
    Replies
    1. I didnt want to say we should stop improving anything... I was just wondering if such restrictions for a potential military use could put pressure on decisions about technical specs. I dont know much about SPAD or the LIDAR like use proposed here. I work on SI-semiconductor handling problems, so my interest is more in 2d imaging (for some problems also in the >1200nm range since SI gets transparent). A 2d sensor with sensitivity both in VIS and >1200nm, with more than VGA resolution, small globalshutter pixels, fit into the usual Cmount-ish formfactors and 29mm-ish camera thermal budgets, output maybe some 500MPix/s over a digital interface that can be fed into a small state of the art FPGA and that is "affordable". I was wondering if production of such a sensor could be prohibited by "non technological" reasons (e.g. export restrictions due to potential military use) even if it would be possible to create. I think currently it is also technologically not possible to create such a sensor, but with emerging technologies mentioned here and in previous posts it seems we are getting closer...

      Delete
  5. This is just a Ge-on-Si process that tsmc has also launched for masses to use. Dont understand the rationale for investment when such a process is available freely now.

    ReplyDelete
    Replies
    1. It is Si only, with some clever way to save Ge.

      Delete
    2. no whatsoever clever way will ever overcome the laws of physics. if detection is based on the inner photoelectric effect, it ends when photons have less energy than the bandgap energy - 1.14eV for SI at 300K, which is equivalent to a photon of 1080nm wavelengh.

      Delete
    3. Si is an indirect bandgap material, so direct relation between bandgap energy and cut-off wavelength does not exist. Photon absorption happens thru intermediate states/phonons.. It can be made sensitive up to 1500nm IMHO, albeit at lower detection performances.

      Delete
    4. my knowledge on this is not very deep, I'd like to learn a bit more about this. could you propose some literature for the basic mechanism? I wonder why then the QE curves of SI based sensors end more or less exactly at the wavelength thats equivalent to the SI bandgap energy. Could you sketch the mechanism that happens when 1500nm photons get detected by a SI photodiode?

      Delete
    5. If you are trying to excite an electron from the valence band to the conduction band, you do need to make up some k from somewhere, like a phonon, due to the indirect bandgap. But if you don't have enough photon energy to make it to the bottom of the conduction band (think E-k) you are going to have little chance of absorption. Strained silicon has a slightly different bandgap. Heavily doped silicon allows some free carrier absorption. With single-crystalline silicon only, a useful 1500nm absorption cutoff is something we'd all like to learn more about. On the other hand, CMOS-compatible structures including nano-particle aka quantum-dot absorption can make CMOS SWIR imaging possible without going to traditional InGaAs or Ge types of material.

      Delete
  6. Thanks Eric! where does this "k" come from? The wikipedia-level reading indicates that in order to shift an electron to conduction band (thats what I thought is the basic mechanism in a photodiode), "a photon and phonon have to be absorbed simultaneously", the photon contributes the energy difference, the phonon the difference in crystal momentum. Can you think of this phonons as some kind of "virtual particles" or "waves" that are a property of SI due to e.g. temperature and its more like a probability issue to make a photon excite an electron since it has to hit it "in the right crystal momentum moment". Or can you think of phonons more like something "externally applied" like the photons? (sorry, I really miss a physics degree... but I plan to get one once I'm retired ;-)

    ReplyDelete
    Replies
    1. Phonons are quantized crystal vibrational waves, and these waves carry momentum. An electron's momentum p = h-bar x k where h-bar is Planck's constant and k, the wave number, is more or less 1/L where L is its wavelength. In silicon, the conduction band minimum energy is offset from the valence band maximum energy by an amount along the k-axis, and an amount (Egap) on the E-axis, on an E-k diagram.

      Another way to think about this (without thinking too hard) is that to break a silicon covalent bond and get a free electron (and hole) requires both energy AND momentum. Photons don't have much momentum so one place the missing momentum can come from is from phonons. In direct bandgap materials you don't need the momentum part to break a covalent bond.

      Delete

All comments are moderated to avoid spam and personal attacks.