Narasimhan said that the new technology may not reach production cars for a decade. Three to four years of research remains, after which automakers might start building it into their vehicles. The technology could double the cost of each headlight, however, at least until economies of scale drive down the manufacturing costs. Ford Motor Co. is said to be interested in it, and the work is being sponsored by Intel Corp., the Office of Naval Research, and the National Science Foundation, among others.
CMU's development is significant because it serves as an alternative to the use of head-up displays to provide better vision in snow and thunderstorms. Up to now, automakers have experimented with computer vision methods, which remove rain and snow streaks from captured images and then project those images onto a head-up display. CMU researchers believe their technique has greater potential because it enables drivers to keep their eyes on the road.
Narasimhan said that he expects the ability to selectively turn on and off pixels to be appealing in other applications. It could enable vehicles to reduce glare for oncoming drivers or intelligently highlight obstacles in the road. "With a million pixels, you can even use the headlights to project images," he said. "You could project arrow marks on the road to tell drivers where to turn, so they don't have to be looking at their GPS all the time."
Very cool development and one that would come in extremely handy in the Northeast where I live. Interesting to see once again software at the heart of driving this kind of new innovation. But waiting a decade to see the technology commercialized--that seems like an awful long time and in not in keeping with today's vastly accelerated time-to-market cycles.
You are correct that the ten year time frame seems long. The tightly coupled embedded systems are available now. Various chip makers have SOCs that can handle this today. On the other hand, it then has to be integrated into the car. Autos are not typically built to be that modular.
Nowadays is relatively easy to integrate external systems in automotive electronics. Even this technological solution could be just reduced in a small compact module placed on its final location at the headlight.
At the moment I think the evolution would be necessary only in the technology of light, for getting compact projectors for high-intensity light (such as current headlight) and low cost, for being a reliable and viable proyect.
I agree with Roberto. It's much easier today to integrate this type of system in current cars. Ten years to too long.
It's definitely a great after market opportunity for an entrepreneur. But, that does bring up the question of Creative Commons. Will it be proprietary or is it considered important enough for general safety to be shared. I would hate to think I'd have to buy a Ford to get these great headlights.
The Digital Light Processing (DLP) technology was invented at Texas Instruments back in 1987, so this application has been technically feasible for 25 years, but just now being demonstrated. Another 10-years to commercialization sounds long, but in the grand scheme of things it sounds about right. Especially if you are talking about a safety device that must be developed and then tested and approved by the NHTSA.
Heck... I'm still waiting for my flying car and they were invented in the 1950s...
Even if the equipment could be just developed in a profitable way, there are a lot of parameters to control and logical problems to solve during the software programing. By the time a solution has been implemented which only works in test enviroment, with homogeneous and static background, and a small deep of falling drops. I think the final solution needs a video signal processing much more complex to deal successfully with real situations, as the great deep of field of falling drops or dinamic and much more random and irregular background. Even the water over the surface of the camera lenses (don´t forget that it's raining!) must be taken into account, getting very difficult to process distorted pictures.
Hopefully that is not just pie in the sky. Such as in other fields, I think though we could have the technology, we still don't know how to use it efficiently and to trasmit a final solution to the market.
Good point about all the gyrations a technology has to go through in terms of being properly vetted by a safety board. Even so, 10 years is such a long time--given the state of technology today, it could be obsolete by then.
Good point about all the gyrations a technology has to go through in terms of being properly vetted by a safety board. Even so, 10 years is such a long time--given the state of technology today, it could be obsolete by then.
I'd worry about the issue of obsolescence also. My TV's DLP has been a source of annoyance to me for years- not quite enough to replace it ($$$!) but enough to have to replace multiple components multiple times. Heat and vibration seem to kill the bulb- is there any of that (heat/vibration) in a car? Plus, if your concern is to "screen out" the water/fog droplets in your vision, if this is a 10-year proposition it seems to me a heads-up-video display is no farther off, and video signal processing is far more advanced than this new idea. I wouldn't expect this to make it in the marketplace - more likely we'll lose the windshield and drive by a large TV screen.
Being a systems designer I tend to look at applications in a different light. Using an LED source of light seems a good idea, but just modulating the light as with a projector seems not so practical. Instead LED color spectrum might be modulated as to provide a light spectrum that has less reflection AND an electronic dash mounted display or heads-up projected image of the road displays a highly processed image of the road ahead. In this way a practical system treating the snow or rain as noise will show the road ahead clearly when the noise is removed by software process. This is how I would design this system, Although retired, I still can be called upon to help out with systems designed to save lives.
Although DLP chips curently have over 2M pixels, the chipset costs over $200 per projector. Besides cost, the key constrain in such a mega-pixel projection headlight would be the brightness. Currently 4000 lumens is considered to be "bright" for a DLP data projector commonly used in conference rooms. A projector headlight for cars would have to be many times brighter than a data projector to illuminate the road ahead for the same distance that current headlights support. Given all the additional heat that will be dissapated from such a bright light source, I don't know if DLP chips could handle it without extensive cooling measures.
My question is what do the drivers of oncoming vehicle see? Currently, the driver is supposed to dim their lights when approaching oncoming traffic, this appears to assume fully illuminating the field of view, which will blind oncoming traffic.
I seriously doubt that this will ever be fielded in any car for cost reasons alone.
I agree Beth, it is very cool. I'm not a snow driver growing up in a place where I saw snow maybe twice as a child.
This project reminds me of many years ago when some cars had the ability to dim their lights automatically. I was a child when it came out but remembered my Dad's cousin had a car with this feature. I remember her saying it would dim it lights when they would pass roadside signs, don't think she kept the car long.
It'll be intersting to see if it catches on. Sometimes ideas like the dim your headlights just really don't catch on and others like power locks and windows do. Perhaps if it becomes a safety issue like the back up cameras it will become legislated.
The concept seems to have a flaw, as the reflection has to take into account the users (driver) eye location. If you move your head, that changes. This would require an eye position sensor and modifcation of the control parameters to account for movement, making calculations even more complex. Between sensing, calculating and controlling there are too many potential failures to achieve robust reliability.
Ultimately the effort and cost would seem to be a lot for a small benefit over just using the low beams and slowing down. If it did manage to work sometimes, it would soon degrade driver skill, and when the inevitable failure happened, the results would be worse than before.
Interesting technology and posts. Some safety concerns: Would drivers with this technology use it cautiously (safely), since they would be able to drive in a blinding thunderstorm at high speeds? Hydroplaning, coming upon slower vehicles at a high rate of speed, etc. I question the use of daytime running lights (DRL's) in poor weather, since many drivers are not properly educated on their use. especially in rain/snow/sleet, etc. The headlights are typically always on, unless a turn signal is activated, and the headlight goes off, until the turn signal is turned off. With DRL's the driver only knows that his dashboard lights are on, assumes he is safe, and the tailights are not on. I consider the lack of tailights, when the DRL's are on to be a safety design failure that needs to be addressed.
I simply can't see this as being a viable option. There are simply too many obstacles in the way of getting the light to where it's needed (pun intended).
Because of the number of raindrops, the projector would need to have multi-megapixels of diodes projecting the light. Pixels that would hit raindrops would be turned off, preventing adequate light from being projected. When it's raining, there will be a raindrop within reasonable range at most every angle, so very few diodes would be on. To compensate, those that do fire will have to be so bright they would burn out your retina if it hit your eye.
Cool idea, but we'll once again see a lot of money spent on something that simply is not practical.
Decades ago, I workled for a small company that had some success in using strobe light effects for the same purpose. he dwell time of any light emission was shorter than the eye's reaction time, so that reflectkions from moving reflectors like snow flakes would be more or less invisible, while reflections from persistent non-moving objects would indeed be perceived. Whether the use of lasers advances this notion is an interesting question.
Upon looking closer at the video, the before-after in the video looks like they just switched from high to low beams. It takes Carnegie-Mellon research to accomplish that? Switch beams? I think I'm underwhelmed. If you need further evidence, look at the volume of light streaks from the illuminated rain toward the bottom of the screen, moving from seconds 4 to 5 in the video: they are undiminished, apart from the dimmed light intensity. You should by rights see fewer light streaks if the tech works- not just a narrower pattern of illumination. Also note the roadsign disappears the moment you hear the click. The middle and edges of the screen see the main reduction- roughly the same as if you switched from higher-aimed high beams to lower-aimed low beams with less intensity. I see this sort of "proof" on YouTube a lot- not exactly a high level of scientific scrutiny. Maybe I could interest the readers in my HHO Generator that improves gas mileage (ha!).
To JMiller, as the comedian Dennis Miller says, "THIN THE HERD!" Darwinism with tough love. You wanna drive faster than it's safe to drive? Fine. Be sure to do so when you're driving along a narrow mountain road with a cliff on one side and a flimsy safety rail.
And, in order to guarantee great results, there should be one item mandated WHEN the auto companies implement these smart headlites. For driving in rain & snow, tires with minimal tread depth SHOULD be required! Then the full effect of driving at increased speed in inclement weather conditions will provide much statistical data for Version 2 of this great idea.
One blogger commented that her dad's cousin had a car w/ auto-dimming headlites. CADILLAC & later LINCOLN had this feature dating back to the mid 1950s. There was a "maic eye" mounted in a pod in the center of the dashboard. On the front was fresnel lens focusered on a light detector. On the rear was a knob inscribed with the words "NEAR" and "FAR". One could adjust the sensitivity of the "auto" function with this knob. In subsequent years, the sensor was moved to various other places on the dashboard. At one point it was nestled in the left corner of the dash.
First of all: WHAT VIDEO? My computer doesn't seem to be able to find it.
Second: This smells like a professor working from some research govt. grant from the dept of transportation.
Third: I'm going to stick my neck out and after admitting that I don't know ANYTHING about relativity other than what I see on TV; I'm going to say this is Bull S**t. I'll predict here that this is impossible. There is no way that even a super computer could account for all the possible angles of reflection between your eye and a rain drop and a headlight bulb no matter how complicated the "bulb" is. What happens if you move your head one inch to the left. What happens if there are more than the 10 or 15 snow flakes that appear in the illustration. What happens if the headlight on car behind you reflect off a flake or drop. What happens if you pass under a street light and it reflects off the proverbial flake or drop.
I'll also predict that this would add millions of dollars to the price of a car.
I give up. When we spend tax dollars (government grant money) for this the whole thing is hopeless. I doubt that the aformentioned professor was working for free in his spare time or that this was funded by a car company (except may be GM).
We need to remember that the people that approved this expenditure are also working for the same "company" that just detained two guys at the Canadian border for trying to illegally cross into the U.S. with contraband egg wtih a toy in it.
I second your third-statement skepticism; Seppanton had that assessment earlier as well.
I also wonder who funds this sort of thing. Hopefully it's from spare change they had lying around, and not interfering with really important research such as whether lab rats prefer light or dark chocolate.
Thr article says: "Demonstration of the prototype system with an artificial rain drop generator is encouraging making the falling rain disappear in front of the observer."
WTF is an "artificial rain drop generator". These guys need to be doing research in a rubber room.
This makes me want to cry. I heard this morning that EVERY american (including my 5 year old God son) already owes $139,000 in national debt and these so-called "researchers" are spending money on this i'm going to try and explain this to Kory (the 5 year old) that his entire life is going to be spent paying this stuff off.
Engineering and science have finally come down to this.
I'm going to work for McDonalds. At least they are not forcing Kory to pay for my damned hamburgers.
I was quite unimressed with the video, but hey, if it's on YouTube it must be right, huh? If the reflection/detection/beam movement is feasible on the average automobile, it will still be some time in the future that this technology is commercially available.
With the technology level required to mask illumination of raindrops, the car would certainly be able to incorporate the technology that masks the light heading toward the oncoming driver's eyes,while leaving the rest of the beam unaffected, so the oncoming driver would see pretty much a standard low beam. Masked High Beam (Glare-Free High Beam) technology already exists, albeit in a fairly simple form, in Europe.
Twenty years ago in my EE senior design class, I proposed an LCD pixel-addressed coating on the windshield to attenuate (not block) the glare from oncoming headlights, to preserve night vision. My advisor told me it was too impractical and automobile manufacturers would never risk it (worst-case failure modes, litigation, etc.) I wonder what he would think of the auto-braking, following distance, drive-by-wire features of today's cars, and what he would say about this.
Really? I just found out I have two $375 headlights on my car and I need special wrenches and a forearm with 2 wrists to install them. I wonder what thiese intelligent headlights would cost. Imagine, just a few short months ago, if the visibility in snow was too short and the reflection was distracting me, I either dimmed the headlights, slowed down, or (heaven forbid) pulled off the road. I actually drove in fog (gasp!) the other night and found the fog lights on my car worked without the headlights. This is another product that will be installed on cars, not because there is a pressing need for improvement, but primarily because it will provide one automaker the opportunity to brag that they have it before any of the others. Surely this will make a future "Designed by Monkeys" column as the thinking world tries to figure out what the designers were thinking.
@bib - I do agree that people's driving skills and knowledge have deteriorated over the years. You don't want to hear my rant about back up cameras and self-parking cars. As I've mentioned here before, I learned how to drive in my Mom's old '68 Mustang. We lived on a hill. With no modesty, I'll say that I'm amazing at parking (have received several standing ovations over the years).
But, I share the road with others. I've been on the Grapevine, a notorious part of I5 just north of LA, on ice and in heavy fog near the top at 4000ft. With less than 100ft visibility, the car next to me turned on his brights. We were all blinded. Including the semi two lanes over. Luckily, no one panicked. He kept them on until we cleared the fog.
Of course, I wanted to drag him out of the car and shred his license. But, lights like these may help with situations like that. The problem exists. A little training would help solve it but realistically, this is the safest solution.
I always seem to bounce back to the question as to how robust the technology is and who will fixt the hardware and software once damged or non-functinal. I will admit that just about everyone now days travels regardless as to how bad weather conditions are. If these headlamps can be made to work properly, there certainly would be value added, possibly as an option on upper-end cars. I can't imagine what the price tag might but I'm sure it would be substantial. Can anyone indicate if there is a maximum speed beyond which the usefulness disappears?
It appears that few if any posters actually read the article and viewed the supporting videos. If one actually reads and looks then one would see how inaccurate the posted comments appear. Can anyone see how you might actually make the idea work and bypass the obvious pitfalls?
It's a very clever idea: and having driven current adaptive high beams that blank out a vertical strip corresponding to oncoming driver's eyes, I think it would be similarly good to drive with.
For this to get on cars on the road, 4 things are needed
1) a massive increase in power for a megapixel projector: current projectors are nowhere near the intensity of headlights (lumen comparison is bogus in this case: think candelas). And putting a beam splitter in the light path will only make it worse.
2) Software that can handle real automotive conditions: when a car is bouncing up and down as well as driving forward and round bends, the predicted droplet path is going to be quite complex! Not to mention that this is a safety critical application.
3) A legal framework that will permit it. Automotive industry is global, so to justify the expenditure you've got to be able to sell in all major markets. That means persuading the relevant authorities to update FMVSS 108, UNECE R123, and Chinese GB-regs for a start: and you won't get far in persuading them unless the problems at 1) and 2) above have been pretty much resolved.
4) A means to build them to an acceptable cost: certainly no automotive manufacturer could make a business case with current megapixel projectors, let alone the super-projectors we need here.
A really bad idea! Most, not all, drivers I encounter have hard enough time driving on dry pavement during the day with all the distractions.....texting (epidemic), eating, chit chatting on the cell phone. Now, lets give these people the ability to drive thier car during a blinding snow or rain storm and still perform all of the above, not to mention that now you will be able to drive your car in conditions that you couldn't even maneuver in to begin with, or the car has the capability to maneuver in. Really dumb. I want my tax money back, thank you.
This is the STUPIDEST design idea I've seen come down the pike in at least an eon!
Here's the ultimate solution! With the 3% to 6% fuel saving that we will all realize with our S/S vehicles, what they SHOULD do is to divert that fuel to a super-efficient Bunsen burner built into the chrome on steel front bumper. When a sensor detects it's either raining and/or snowing, a spark-gap ignitor initializes the atomized fuel vapor to expel a plasma-hot flame in front of the vehicle. The rain/snow will be dispersed, and driving safety will be enhanced. By eliminating the water build-up, the roads will stay drier & safer to travel on.
The only problem I foresee is that the nozzles built into the bumper would have to be manufactured of an alumina substrate to withstand the plasma temperatures expected.
A far more cost effective way of improving headlights would be to ensure that the bulb, be it incandescent or an LED assemly can be changed easily and effectively, so we do not have one in three cars (my estimate) with faulty headlights, largely because the bulbs cannot be changed without major disassembly - I have even heard of one car that needs the bumper removed to change a headlight bulb!
A reversion to TRUE foglights which have an accurate cut-off of the top of the beam, instead of 'intense visililty lights' currently fitted and misdescribed as foglights could be coupled with one relatively simple improvement. The switching, now likely to be mostly electronic, should switch between NORMAL and FOG/SNOW; the dipping on Normal should be as current main lighting, but dipping when FOG/SNOW selected being between DIPPED BEAM and (true) FOGLIGHTS.
This will overcome the fact so few know how to use foglights, and virtually all factory-fitted foglights being rubbish.
This may be just fine for snow and rain. But what happens if the face of the headlight freezes up. 100 watts of bulbs inside the headlight will take care of that. Would high tech headlights, that like to be kept cool, do that.
Transfers the control of a large number of motion axes from one numerical control kernel to another within a CNC system, using multiple NCKs, and enables implement control schemes for virtually any type of machine tool.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.