Science Diet: Google Drives Into the Future

By Nithyanand Rao and Raghavi Kodati

1. Google’s self-driving cars 

A century and more ago, a “computer” referred to a human being, usually a woman, who was assigned routine computational work during wartime or as part of scientific research projects. Today of course, it refers to machines. With recent developments, the term “driver” could, in a few years, refer to cars instead of humans.

ScienceDietGoogle has been working on self-driving cars — modified cars fitted with cameras, lasers and other sensors — for the last five years. (IEEE Spectrum has a Special Report on self-driving cars here.) Though autonomous, these cars could, if required, hand over control to a human. Now though, Google has gone a step further. Instead of hacking existing vehicles, they’ve built their own car — without brakes, accelerators or a steering wheel; in other words, completely autonomous. The car is electric-powered, with a range of 100 miles and a speed limited to 25 miles per hour. With the driver taken out of the picture completely, the car can be summoned and its destination specified using a smartphone app. And it works so well that John Markoff of the New York Times called it “boring.”

A prototype of Google's self-driving car.
A prototype of Google’s self-driving car. Credit: Google.

How does it work? Data. As mentioned in a previous edition of Science Diet, one approach to artificial intelligence is to resort to brute-force, by collecting massive amounts of data and using computational power to solve otherwise intractable problems. Google’s approach has been to create a virtual representation of the real world environment the cars have to navigate, including details like the position and height of traffic lights and curbs. This data is then fed into the car’s memory before the trip, telling the car what the world looks like when it is devoid of other cars and people. This makes the car’s job easier — to navigate the real world, it first needs to figure out its precise position, and then find out how the real world is different from the idealized world in its memory.

Naturally, Google is coy about its vision behind all this work. What they’ve let on is that they’re targeting people who can’t afford cars right now. They hope to make available a fleet of such autonomous vehicles for city transport.

Lawmakers in California have already come up with regulations for the testing of autonomous vehicles. Although it hasn’t happened so far, what happens in the event of such a car, in spite of all the clever algorithms, violating a traffic rule? The person sitting in what is nominally the driver’s seat is no longer in control. So who pays the fine? The programmer who wrote the code that made the mistake? At this experimental stage, Google takes the responsibility.

2. Electrifying rural India with microgrids

About a quarter of our country’s population does not have access to electricity, being unconnected to the power grid. Several startup companies have found a solution, with the so-called microgrids. These are powered by solar panels, and in some places, hydropower or biomass in the form of rice husk. Just like mobile phones bypassed landlines, these entrepreneurs hope that their microgrids will enable people living in rural areas to bypass the power grid altogether, as a long-term solution.

These microgrids have the advantage of being a quickly implementable solution, of being low-carbon or carbon-free, and can be scaled up faster than traditional power lines. But there are problems: not enough sunlight during winters, and the rising demand and consequently the rising prices of once-unwanted rice husk. A bigger difficulty is in getting farmers with irregular incomes to pay regular bills. One solution is to have modified meters that work like prepaid cellphones, doling out electricity in chunks.

Then there are some not-so-obvious difficulties. Such as “higher caste” folks not wanting their electricity lines to be routed through the houses of “lower caste” people. Some things never change.

3. How much water do you eat?

That’s right. Not drink, but eat.

We need water for cooking food, washing clothes, for bathing and for use in our toilets. That’s the usage which is obvious and visible. But then there’s the invisible part: all the various goods we use – paper, clothes, plastics – invariably require water during their production. Over and above that, water is used in huge quantities to produce the food we eat. In particular, if you eat lots of meat, a staggering 92% of your water usage can be traced back to the production of this meat. The average vegetarian consumes only half this amount of water in the food he or she eats. But there’s more: a kilogram of coffee beans, for instance, requires 18900 litres of water to produce. Rice is far cheaper, coming in at 2500 litres. Find out more, here.

4. Poisoned light

Compact Fluorescent Lamp (CFLs) are rightly popular for their lower electricity consumption. In fact, CFL production in India increased from 67 million in 2005 to 304 million in 2010. But what happens when a CFL reaches the end of its life? You’d probably throw them in the garbage and it would eventually end up at landfills. The ragpickers who handle these, often children, are in danger. Because CFLs contain mercury, which can adversely affect the brain and the nervous system. Improper disposal of CFLs can lead to mercury entering water bodies or the food chain.

However, a survey conducted revealed that 95% respondents were unaware of the presence of mercury in CFLs. The CFL manufacturers do not label their products with information on how much mercury they contain and there are no regulations in India on permissible levels of mercury in CFLs. Germany, for instance, permits 5 mg of mercury in CFLs, while in India it is found to be as high as 21 mg.

Says an associate director of one of the NGOs working in this area: “Most of the CFL manufacturers are multinationals. If they can adhere to mercury limit for lamps in European nations, why can’t they do it here?”

5. How do you divide a cake in an envy-free, equitable, and efficient way?

Division of assets is a routine and common problem – division of toys among children, division of property between siblings, division of business between partners, etc. The most intuitive method of executing a division is where one of the stakeholders proposes a cut, and the other one chooses one of the parts. Although this method ensures the division is envy-free, it may not be efficient; that is, each of the stakeholders might not end up with the parts that they value the most. Erica Klarreich writes about algorithms that perform an efficient and envy-free division, and details the applications of these algorithms to complex situations. Whoever said math wasn’t useful?

Write a Comment

Your email address will not be published. Required fields are marked *