Fog.

The second a driver hears that word is the second s/he knows that life just got more complicated. Those 25, 50, 100 car pileups almost always have fog as their underlying cause. (The other exception- ice.)

Back in the early 1970s, cars routinely piled up on what the world thought was Walton Mountain. (Unless, of course, one lived in or near the University of Virginia. Because the name of the mountain was probably Afton Mountain.) And, Interstate 64 connected Virginia Beach, Newport News, Richmond, Charlottesville, and Staunton. A cross-road that traversed two of the busiest highways in the east- Interstates 95 and 81.

Being among the Blue Ridge Mountains, I 64 routinely was masked by fog. Not surprisingly, the same road was often littered with car wrecks and bodies. Because the pea-soup fog rendered visibility to be zero. And, that might have been less of a problem if one could warn vehicles that the fog was rolling in.

But, this fog was not really related to any weather conditions. One could be driving on I-64 under perfectly clear, perfectly dry road conditions. And, then ascend the mountain to be enveloped with fog. With no notice, which meant the vehicles were traveling at 65, 75, or even 85 miles an hour- and would clunk into cars stopped dead, or meandering at 8 mph.

A solution was found. No, not to the fog. But, to alert the drivers. After all, fog is simply a colloid. A colloid is a homogenous solution, where particles of one material are fully dissolved in another. Homogenized milk is a colloid. So is chocolate pudding. And, fog is a colloid- where water vapor is fully dissolved in air.

A feature of colloids is the Tyndall effect. That effect describes how light is dispersed, deflected, or scattered when it travels through a colloidal solution. Which explains why,  if a focused light were to be projected from one side of the road, the light would be fully received by a receptor on the other side of the road, most of the time. Up until the moment fog begins to form- and then the light gets scattered. So, the receptor (which would no longer be receiving the photons of light expected) would send a signal to a flashing yellow light (at this point in time, road visibility would be less 1400 feet), denoting the impending (as well as continued) appearance of fog. Thereby alerting the drivers of the changed conditions.  Telling them to slow down.

Some 843 incandescent bulbs were all it took. (Some 30 years later those bulbs were replaced by LED’s, using much less power. And, less likely to be destroyed by passing vehicles.)

Now, just consider how a foggy situation would present itself to an autonomous vehicle. Which controls its actions with between 5 and 10 cameras. And, LIDAR (light detection and ranging; sometimes the L is considered to be laser instead of light). Just from their names, you recognize that these devices rely on light transmission. Oh, sure, the front bumpers have (what they call 360 degree) radar. Yup. Those vehicles will be stymied in fog.

Well, maybe not. You see, the MIT Media Lab is developing a solution. Based upon computational photography. Developed by Guy Satat (whose advisor is Dr. Ramesh Raskar), who hopes to obtain his PhD from MIT soon enough. (Probably when he extends this concept).

Basically, the research currently employs a small tank of water, with a humidifier and vibrating motor. Once it creates the ‘research’ fog, visible is limited to 36 cm. (That is way thicker than regular fog, where visibility is defined as under 3300 feet (1 kilometer). And, with their solution, visibility would be extended to 57 cm. (The MIT Media Lab feels the scaled up version would extend visibility in fog  to between 30 to 50 meters. Which means a very safe stopping distance if the autonomous vehicle were traveling between 20 and 30 mph.

What is the basis of their solution? A time-of-flight camera, employing ultrashort bursts of laser beams, and analyzing the time it takes the light to move through the colloid. Under normal conditions, the circuit time indicates the actual distance. Once dispersed by fog, the circuit times are vastly different. Based upon fog density, the arrival times of the reflected light adhere to gamma distributions. Each of the 1024 pixels on the sensor create their own gamma distribution (more complex than the normal “Gaussian” distribution) and the camera counts the number of light particles (called photons) every 56 picoseconds (i.e., 56 trillionth of a second)- and compares the light received to the expected distribution. And, those deviations correlate with physical obstacles.

Et, voila.

Now, if only the pedestrians would keep to their assigned paths and not run in front of vehicles. That still seems like the next obstacle to be conquered to ensure safe people AND safe autonomous automobile travel.

Today is the the 5th day of Passover (Pesach), the holiday of Freedom.  And, it is also the 50th anniversary of the assassination of the Reverend Martin Luther King.   And, while (thankfully) no one has heard from David Duke in a while, he was busy insulting folks, including my rabbi and my synagogue.  Which is why we made out pilgrimage to Selma and Montgomery last year.

You can learn more about the reasons and the trip here.

Paperback edition

Kindle edition