This is what sunset looks like on Mars
The Curiosity rover's high-resolution camera has returned pictures of a dusty sunset on the Red Planet.
The Curiosity rover has been sending back data and photos since it landed on Mars nearly three years ago. Unsurprisingly, there are a lot of rocks in those images.
But finally, for the first time, on Curiosity's 956th solar Mars day -- April 15, 2015, to all you Earthlings out there -- its Mastcam high-resolution, colour camera captured the setting sun on the Red Planet. (A solar Mars day, or sol, by the way, lasts just over 24 hours and 39 minutes.)
Mars' sunset of cool blues looks starkly different from our warm-hued equivalent. This is because of the thinner Martian atmosphere and the dust in it. The images of the sunset were taken in between dust storms, but with plenty of dust still in the air. This allowed researchers to analyse the vertical distribution of dust in the Martian atmosphere.
"The colours come from the fact that the very fine dust is the right size so that blue light penetrates the atmosphere slightly more efficiently," Curiosity science team member Mark Lemmon of Texas A&M University, who planned the observations, said Thursday in a statement from NASA.
"When the blue light scatters off the dust, it stays closer to the direction of the sun than light of other colours does. The rest of the sky is yellow to orange, as yellow and red light scatter all over the sky instead of being absorbed or staying close to the sun."
That's the opposite of Earth sunsets, where the sky grows red and yellow as the sun dips into the horizon, with the rest remaining blue.
The images were returned from Curiosity in black and white. If you look at them carefully, you'll see a sort of hashed texture to the image. This pattern is called the Bayer matrix, and it's how the Mastcam (and other digital cameras) codes colour into images.
The Bayer filter is a series of colour filters painted over the Mastcam's detector. Each pixel is painted either red, green or blue, and when the Mastcam's shutter opens, light enters into each of these pixels according to how the human eye would see them.
However, the sensor itself can't interpret the colours on its own. Digital cameras have on-board software to decode the greyscale matrix that emerges, but some of the Bayer data coming back from Curiosity needs to be decoded on Earth. This is how we are able to extrapolate colour data from what looks like a black and white image.
These images, taken over a period of about 6 minutes, 51 seconds, were decoded by space imaging enthusiast Damia Bouic, who said in her blog last week:
"These images came in black and white, with a Bayer matrix. I had to process them through Gimp and G'MIC in order to rebuild the colours. A little processing in Pixelmator to remove these ugly white stripes due to pixels overloading. A little bit of denoising to remove Bayer matrix artifacts, et voilà!"
Although it's in glorious high-resolution, and the first view from Curiosity, this is not the first Mars sunset we've ever seen. In 2010, NASA's Jet Propulsion Laboratory released a video of the Martian sunset as observed by the Opportunity rover on November 4 and 5, 2010.
THIS WEEK'S MUST READS
How a French vacation shows we need high-tech addresses
Even in the Net era, physical addresses are important in mature and developing economies. That's why companies like Google and What3words offer global location-encoding systems.
CARGESE, Corsica, France -- Ours is a 21st-century family. On our April vacation to Corsica, we booked lodging through Airbnb, navigated by satellite and followed trail maps on a smartphone.
But in one way, our two-week journey to France's scenic Mediterranean isle was like being dragged back to the Middle Ages: Corsican dwellings often lack street addresses. Each of the four times we moved to a new house, we wasted lots of time trying to figure out where the heck to go. Apartment buildings in mainland France don't number their apartments, and postal codes cover areas too big to be really useful, but Corsica was way worse.
There's good news, though. Everyone from mapping goliath Google to small startups like What3wordsare bringing new address technology that's trying to fix the situation. Even better, that technology will help countless people in developing economies who lack functioning address systems -- not just tourists with first world problems.
That could be a big deal. Four billion people lack addresses, according to the United Nations Development Program. That makes it hard for them to receive packages, tap into government benefits, sign up for electricity and water utilities and get help from emergency services. Even in our Internet-infused future, physical addresses matter.
Corsican tirade
Corsica, birthplace of Napoleon Bonaparte and for centuries a Genoan outpost, is robbed of modern efficiencies by its imperfect addressing.
To find one house, our instructions were to drive 300 meters past a particular intersection, look for a bus stop, then for a nearby metal gate. For another, we met the proprietor at a nearby cafe. For a third, we had to find a particular concrete-paved road, look for a line of garages, drive past three buildings and then look for another that had the words "BAT G" in red -- lettering an inch high on an electrical junction box, as it turned out. The fourth was 7km from one town and 5km from another. We found it easily -- but only because I'd spent about half an hour in advance scrutinizing Google Maps satellite photos.
I was flabbergasted. France is the country that invented the metric system and Cartesian coordinates! Paris was the runner-up for the location of the Prime Meridian, where longitude is zero. Baron Haussmann replaced a maze of medieval streets in the country's capital with rationally planned boulevards. This is the nation that hasn't figured out street addresses yet?
But it turns out it's not alone. The Republic of Ireland isn't even scheduled to get its first postal code system, Eircode, until this summer. Nigeria and Lebanon began nationwide address systems in 2013. Many Japanese streets don't have names, and some address numbers are assigned not by position but by the date the building was registered. Even where real-world addresses exist, it's very hard for programmers at shipping and mapping companies to systematically handle all their irregularities.
This kind of confusion made me realize how I'd taken US street numbers and ZIP codes for granted. And it gave me an even better appreciation for the UK's postal codes, which are precise enough to punch in to Google Maps or a car's sat-nav system and which are deeply embedded into the country's government and commercial services.
New technology
Where I see problems, businesses see opportunities.
On my second-to-last day in Corsica, Google publicly launched a project called Open Location Codethat converts long numeric latitude-longitude coordinates into a 10-digit sequence of numbers and letters. Napoleon's birthplace in Ajaccio, Corsica, for example, is at 8FHCWP9Q+6C. Adding human-memorable elements can shorten the address, though; 3F88+9M Albuquerque will get you to the National Museum of Nuclear Science and History.
Google hopes its codes will help with navigation and package delivery.
"People are currently using addresses like 'behind the old bus stop' that either presuppose detailed local knowledge or are more like directions," said Doug Rinckes, a technical program manager at Google Switzerland. "If we can give people an alternative that means that DHL, TNT or UPS are even able to try to deliver something to you in places like unmapped towns, or cities with unnamed streets, that would be amazing."
Google has released its Open Location Code algorithm as open-source software so anybody can use it for free, but the most obvious candidate is Google Maps, a service embedded in millions of Android phones. Open Location Code support in Google Maps could ease tasks like telling your friend where a trailhead is located -- or make it easier for Airbnb to help me bookmark the location of my rented house. The Chez Joelle guesthouse has no street address besides just "Le Casone" in the Corsican village of Ota, but with Open Location Code you can find it at 8FJC7P5V+98 or 5V+98 Ota, France.
Open Location Code's designers hope for Google Maps support. "That's definitely my hope (and I'm working with colleagues)," Rinckes said in a mailing list comment. "We have been asked why we open-sourced it first, and the reason was that we wanted to get feedback early on and see if the algorithm survived on its own merits, or if it needed to be modified."
Google isn't alone. The MapCode Foundation, run by the co-founder of Netherlands-based navigation specialist TomTom International among others, has an alphanumeric system for precise locations, too. NAC Geographic Products has licensed its system to some countries. In India and other emerging markets, a company called Zippr hopes to ease package delivery and navigation with its smartphone-based addressing system.
More memorable
A UK-based startup called What3words has a different approach, though. Its algorithm divides the world into 57 trillion squares of land 3 meters on a side, each with its own three-word label. Times Square in central New York City is bolts.native.year and Faneuil Hall in Boston is atomic.camera.poem, for example.
"The reason we're better than the alphanumerics is memorability," said Giles Rhys Jones, marketing chief of the eight-person company. "Your ability to remember three words is nearly perfect. Your ability to remember anything over seven digits or alphanumeric characters is zero."
The company still uses an algorithm to generate the three-word combinations, which means the technology works even without a network connection. The system works with an expanding list of non-English languages, too, and is optimized so areas with higher population density get shorter names, Rhys Jones said. For example, Mount Rushmore National Memorial in South Dakota is occasionnellement.antiaérienne.abrègement in French but the Eiffel Tower in Paris is galon.baisser.coupe.
What3words hopes to make money through use of its service, with big customers paying for access to its application programming interface.
Finding national monuments is fine, but the company's business hinges more on adoption in developing markets. "We're talking to a couple governments in Africa to put our system in place as the postal system. They have nothing there," Rhys Jones said.
For example, in the Brazilian slums known as favelas, one entrepreneur is using What3words to operate the last leg of a delivery service in the chaos of unnamed streets.
And that's where, like me finding a business faster in Corsica, the world stands to gain.
"It used to be you were unconnected and unbanked," but smartphones and electronic banking services are coming to developing nations, Rhys Jones said. "We think unaddressed is the third part you really need for social and economic development."
THIS WEEK'S MUST READS
LATEST GALLERIES FROM CNET
When self-driving cars go bump in the street
Technically Incorrect: A smattering of the self-driving cars on California roads have been involved in incidents in the last six months. None of them, say their manufacturers, were the fault of the cars.
Cars don't cause accidents. People cause accidents.
This is the hope and dream (and, surely, motto) of those involved in designing self-driving cars.
The whole idea is built on the nerds' rational nirvana in which incidents below perfection must be the fault of the imperfect human.
I'm wondering about this because of an AP reportthat declares four self-driving cars have been involved in accidents in California since September. I couldn't decide whether this was an important or insignificant proportion, until I discovered that there had been around 50 self-driving cars on California's roads during that time.
A quick mustering of the math tells me that these incident-affected cars represent around 8 percent of all the the self-driving cars on the road.
I fancy that's a significant number, but also one that may not prove anything.
Three of the cars belonged to Google and one to Delphi, the auto parts company that in March set one of its self-driving cars to drive itself across America.
I contacted both companies to ask about the circumstances of these incidents.
A Delphi spokeswoman told me: "We have not had any accidents involving any of our test vehicles while in automated mode. There was, however, an accident in October 2014 involving one Delphi automated vehicle [that] had nothing to do with automated driving. While operating in manual mode and stopped at an intersection, our car was struck by another vehicle that traveled across the median. A police report indicates the fault of the accident is with the second vehicle. Not Delphi. A police report filed in Mountain View confirms this."
Google referred me to a new post from Chris Urmson, director of its self-driving program. He insisted that Google's cars have a much better reaction time than do humans.
And he cited statistics, of course: "Over the 6 years since we started the project, we've been involved in 11 minor accidents (light damage, no injuries) during those 1.7 million miles of autonomous and manual driving with our safety drivers behind the wheel, and not once was the self-driving car the cause of the accident."
Perfection is a beautiful thing for an engineer.
He who hesitates is bumped
Urmson said that Google's self-driving cars have been hit seven times from behind, as well as a higher number of routine incidents when driving in cities.
He explained that the company's cars are programmed to pause a little longer at intersections, because people behave especially badly there. In essence, his post smacks of an engineer's frustration that human beings are so maddeningly unpredictable.
A Google spokeswoman clarified that of the 1.7 million miles Urmson referred to, almost 1 million were while a vehicle was in autonomous mode.
Looking at the high proportion of self-driving cars involved in accidents may not be helpful, as so many minor accidents simply aren't reported at all by clumsy, embarrassed and insurance-dodging humans. Google, by contrast, must disclose all accidents to the DMV.
By the way, none of Google's self-driving cars goes on the road without a driver in it.
Clearly, as self-driving car technology is thrust headlong into our lives, there will be a gestation period in which more accidents will happen. Those at or near the controls of self-driving cars will make mistakes. There will surely be technological failures too.
Google not only trumpets the safety of its cars, but also the idea of greater efficiency, leading to less personal car ownership, fewer deaths and traffic jams, and a diminution in the number of parking lots.
Less often discussed is the ultimate goal of ensuring that humans never drive again. Indeed, the ever ambitious Elon Musk predicted recently that self-driving cars will be the norm in 20 years time.
We have no idea how this might feel, because we have no idea how much and how rapidly technological change will affect us even in the near future.
Once upon a time we talked on something called a phone. Now we take calls on our watches, even though it looks ridiculous.
The more we rely on the likes of Uber for getting around, the more the mere notion of driving being exhilarating may fade.
Then we'll be happy to be passengers, hoping that technology carries us safely to our destinations. Because technology is smarter than we are, isn't it?
No comments:
Post a Comment
Please leave a comment-- or suggestions, particularly of topics and places you'd like to see covered