Tag Archives: NVIDIA

#432878 Chinese Port Goes Full Robot With ...

By the end of 2018, something will be very different about the harbor area in the northern Chinese city of Caofeidian. If you were to visit, the whirring cranes and tractors driving containers to and fro would be the only things in sight.

Caofeidian is set to become the world’s first fully autonomous harbor by the end of the year. The US-Chinese startup TuSimple, a specialist in developing self-driving trucks, will replace human-driven terminal tractor-trucks with 20 self-driving models. A separate company handles crane automation, and a central control system will coordinate the movements of both.

According to Robert Brown, Director of Public Affairs at TuSimple, the project could quickly transform into a much wider trend. “The potential for automating systems in harbors and ports is staggering when considering the number of deep-water and inland ports around the world. At the same time, the closed, controlled nature of a port environment makes it a perfect proving ground for autonomous truck technology,” he said.

Going Global
The autonomous cranes and trucks have a big task ahead of them. Caofeidian currently processes around 300,000 TEU containers a year. Even if you were dealing with Lego bricks, that number of units would get you a decent-sized cathedral or a 22-foot-long aircraft carrier. For any maritime fans—or people who enjoy the moving of heavy objects—TEU stands for twenty-foot equivalent unit. It is the industry standard for containers. A TEU equals an 8-foot (2.43 meter) wide, 8.5-foot (2.59 meter) high, and 20-foot (6.06 meter) long container.

While impressive, the Caofeidian number pales in comparison with the biggest global ports like Shanghai, Singapore, Busan, or Rotterdam. For example, 2017 saw more than 40 million TEU moved through Shanghai port facilities.

Self-driving container vehicles have been trialled elsewhere, including in Yangshan, close to Shanghai, and Rotterdam. Qingdao New Qianwan Container Terminal in China recently laid claim to being the first fully automated terminal in Asia.

The potential for efficiencies has many ports interested in automation. Qingdao said its systems allow the terminal to operate in complete darkness and have reduced labor costs by 70 percent while increasing efficiency by 30 percent. In some cases, the number of workers needed to unload a cargo ship has gone from 60 to 9.

TuSimple says it is in negotiations with several other ports and also sees potential in related logistics-heavy fields.

Stable Testing Ground
For autonomous vehicles, ports seem like a perfect testing ground. They are restricted, confined areas with few to no pedestrians where operating speeds are limited. The predictability makes it unlike, say, city driving.

Robert Brown describes it as an ideal setting for the first adaptation of TuSimple’s technology. The company, which, amongst others, is backed by chipmaker Nvidia, have been retrofitting existing vehicles from Shaanxi Automobile Group with sensors and technology.

At the same time, it is running open road tests in Arizona and China of its Class 8 Level 4 autonomous trucks.

The Camera Approach
Dozens of autonomous truck startups are reported to have launched in China over the past two years. In other countries the situation is much the same, as the race for the future of goods transportation heats up. Startup companies like Embark, Einride, Starsky Robotics, and Drive.ai are just a few of the names in the space. They are facing competition from the likes of Tesla, Daimler, VW, Uber’s Otto subsidiary, and in March, Waymo announced it too was getting into the truck race.

Compared to many of its competitors, TuSimple’s autonomous driving system is based on a different approach. Instead of laser-based radar (LIDAR), TuSimple primarily uses cameras to gather data about its surroundings. Currently, the company uses ten cameras, including forward-facing, backward-facing, and wide-lens. Together, they produce the 360-degree “God View” of the vehicle’s surroundings, which is interpreted by the onboard autonomous driving systems.

Each camera gathers information at 30 frames a second. Millimeter wave radar is used as a secondary sensor. In total, the vehicles generate what Robert Brown describes with a laugh as “almost too much” data about its surroundings and is accurate beyond 300 meters in locating and identifying objects. This includes objects that have given LIDAR problems, such as black vehicles.

Another advantage is price. Companies often loathe revealing exact amounts, but Tesla has gone as far as to say that the ‘expected’ price of its autonomous truck will be from $150,0000 and upwards. While unconfirmed, TuSimple’s retrofitted, camera-based solution is thought to cost around $20,000.

Image Credit: chinahbzyg / Shutterstock.com Continue reading

Posted in Human Robots

#432456 This Planned Solar Farm in Saudi Arabia ...

Right now it only exists on paper, in the form of a memorandum of understanding. But if constructed, the newly-announced solar photovoltaic project in Saudi Arabia would break an astonishing array of records. It’s larger than any solar project currently planned by a factor of 100. When completed, nominally in 2030, it would have a capacity of an astonishing 200 gigawatts (GW). The project is backed by Softbank Group and Saudi Arabia’s new crown prince, Mohammed Bin Salman, and was announced in New York on March 27.

The Tengger Desert Solar Park in China, affectionately known as the “Great Wall of Solar,” is the world’s largest operating solar farm, with a capacity of 1.5 GW. Larger farms are under construction, including the Westlands Solar Park, which plans to finish with 2.7 GW of capacity. But even those that are only in the planning phases are dwarfed by the Saudi project; two early-stage solar parks will have capacity of 7.2 GW, and the plan involves them generating electricity as early as next year.

It makes more sense to compare to slightly larger projects, like nations, or even planets. Saudi Arabia’s current electricity generation capacity is 77 GW. This project would almost triple it. The current total solar photovoltaic generation capacity installed worldwide is 303 GW. In other words, this single solar farm would account for a similar installed capacity as the entire world’s capacity in 2015, and over a thousand times more than we had in 2000.

That’s exponential growth for you, folks.

Of course, practically doubling the world’s solar capacity doesn’t come cheap; the nominal estimate for the budget is around $200 billion (compared to $20 billion for around half a gigawatt of fusion, though, it may not seem so bad.) But the project would help solve a number of pressing problems for Saudi Arabia.

For a start, solar power works well in the desert. The irradiance is high, you have plenty of empty space, and peak demand is driven by air conditioning in the cities and so corresponds with peak supply. Even if oil companies might seem blasé about the global supply of oil running out, individual countries are aware that their own reserves won’t last forever, and they don’t want to miss the energy transition. The country’s Vision 2030 project aims to diversify its heavily oil-dependent economy by that year. If they can construct solar farms on this scale, alongside the $80 billion the government plans to spend on a fleet of nuclear reactors, it seems logical to export that power to other countries in the region, especially given the amount of energy storage that would be required otherwise.

We’ve already discussed a large-scale project to build solar panels in the desert then export the electricity: the DESERTEC initiative in the Sahara. Although DESERTEC planned a range of different demonstration plants on scales of around 500 MW, its ultimate ambition was to “provide 20 percent of Europe’s electricity by 2050.” It seems that this project is similar in scale to what they were planning. Weaning ourselves off fossil fuels is going to be incredibly difficult. Only large-scale nuclear, wind, or solar can really supply the world’s energy needs if consumption is anything like what it is today; in all likelihood, we’ll need a combination of all three.

To make a sizeable contribution to that effort, the renewable projects have to be truly epic in scale. The planned 2 GW solar park at Bulli Creek in Australia would cover 5 square kilometers, so it’s not unreasonable to suggest that, across many farms, this project could cover around 500 square kilometers—around the size of Chicago.

It will come as no surprise that Softbank is involved in this project. The founder, Masayoshi Son, is well-known for large-scale “visionary” investments. This is suggested by the name of his $100 billion VC fund, the Softbank Vision Fund, and the focus of its investments. It has invested millions of dollars in tech companies like Uber, IoT, NVIDIA and ARM, and startups across fields like VR, agritech, and AI.

Of course, Softbank is also the company that bought infamous robot-makers Boston Dynamics from Google when their not-at-all-sinister “Project Replicant” was sidelined. Softbank is famous in Japan in part due to their mascot, Pepper, which is probably the most widespread humanoid robot on the planet. Suffice it to say that Softbank is keen to be a part of any technological development, and they’re not afraid of projects that are truly vast in scope.

Since the Fukushima disaster in 2011 led Japan to turn away from nuclear power, Son has also been focused on green electricity, floating the idea of an Asia Super Grid. Similar to DESERTEC, it aims to get around the main issues with renewable energy (the land use and the intermittency of supply) with a vast super-grid that would connect Mongolia, India, Japan, China, Russia, and South Korea with high-voltage DC power cables. “Since this is such a grandiose project, many people told me it is crazy,” Son said. “They said it is impossible both economically and politically.” The first stage of the project, a demonstration wind farm of 50 megawatts in Mongolia, began operating in October of last year.

Given that Saudi Arabia put up $45 billion of the Vision Fund, it’s also not surprising to see the location of the project; Softbank reportedly had plans to invest $25 billion of the Vision Fund in Saudi Arabia, and $1 billion will be spent on the first solar farms there. Prince Mohammed Bin Salman, 32, who recently consolidated power, is looking to be seen on the global stage as a modernizer. He was effusive about the project. “It’s a huge step in human history,” he said. “It’s bold, risky, and we hope we succeed doing that.”

It is the risk that will keep renewable energy enthusiasts concerned.

Every visionary plan contains the potential for immense disappointment. As yet, the Asian Super Grid and the Saudi power plan are more or less at the conceptual stage. The fact that a memorandum of understanding exists between the Saudi government and Softbank is no guarantee that it will ever be built. Some analysts in the industry are a little skeptical.

“It’s an unprecedented construction effort; it’s an unprecedented financing effort,” said Benjamin Attia, a global solar analyst for Green Tech Media Research. “But there are so many questions, so few details, and a lot of headwinds, like grid instability, the availability of commercial debt, construction, and logistics challenges.”

We have already seen with the DESERTEC initiative that these vast-scale renewable energy projects can fail, despite immense enthusiasm. They are not easy to accomplish. But in a world without fossil fuels, they will be required. This project could be a flagship example for how to run a country on renewable energy—or another example of grand designs and good intentions. We’ll have to wait to find out which.

Image Credit: Love Silhouette / Shutterstock.com Continue reading

Posted in Human Robots

#432165 Silicon Valley Is Winning the Race to ...

Henry Ford didn’t invent the motor car. The late 1800s saw a flurry of innovation by hundreds of companies battling to deliver on the promise of fast, efficient and reasonably-priced mechanical transportation. Ford later came to dominate the industry thanks to the development of the moving assembly line.

Today, the sector is poised for another breakthrough with the advent of cars that drive themselves. But unlike the original wave of automobile innovation, the race for supremacy in autonomous vehicles is concentrated among a few corporate giants. So who is set to dominate this time?

I’ve analyzed six companies we think are leading the race to build the first truly driverless car. Three of these—General Motors, Ford, and Volkswagen—come from the existing car industry and need to integrate self-driving technology into their existing fleet of mass-produced vehicles. The other three—Tesla, Uber, and Waymo (owned by the same company as Google)—are newcomers from the digital technology world of Silicon Valley and have to build a mass manufacturing capability.

While it’s impossible to know all the developments at any given time, we have tracked investments, strategic partnerships, and official press releases to learn more about what’s happening behind the scenes. The car industry typically rates self-driving technology on a scale from Level 0 (no automation) to Level 5 (full automation). We’ve assessed where each company is now and estimated how far they are from reaching the top level. Here’s how we think each player is performing.

Volkswagen
Volkswagen has invested in taxi-hailing app Gett and partnered with chip-maker Nvidia to develop an artificial intelligence co-pilot for its cars. In 2018, the VW Group is set to release the Audi A8, the first production vehicle that reaches Level 3 on the scale, “conditional driving automation.” This means the car’s computer will handle all driving functions, but a human has to be ready to take over if necessary.

Ford
Ford already sells cars with a Level 2 autopilot, “partial driving automation.” This means one or more aspects of driving are controlled by a computer based on information about the environment, for example combined cruise control and lane centering. Alongside other investments, the company has put $1 billion into Argo AI, an artificial intelligence company for self-driving vehicles. Following a trial to test pizza delivery using autonomous vehicles, Ford is now testing Level 4 cars on public roads. These feature “high automation,” where the car can drive entirely on its own but not in certain conditions such as when the road surface is poor or the weather is bad.

General Motors
GM also sells vehicles with Level 2 automation but, after buying Silicon Valley startup Cruise Automation in 2016, now plans to launch the first mass-production-ready Level 5 autonomy vehicle that drives completely on its own by 2019. The Cruise AV will have no steering wheel or pedals to allow a human to take over and be part of a large fleet of driverless taxis the company plans to operate in big cities. But crucially the company hasn’t yet secured permission to test the car on public roads.

Waymo (Google)

Waymo Level 5 testing. Image Credit: Waymo

Founded as a special project in 2009, Waymo separated from Google (though they’re both owned by the same parent firm, Alphabet) in 2016. Though it has never made, sold, or operated a car on a commercial basis, Waymo has created test vehicles that have clocked more than 4 million miles without human drivers as of November 2017. Waymo tested its Level 5 car, “Firefly,” between 2015 and 2017 but then decided to focus on hardware that could be installed in other manufacturers’ vehicles, starting with the Chrysler Pacifica.

Uber
The taxi-hailing app maker Uber has been testing autonomous cars on the streets of Pittsburgh since 2016, always with an employee behind the wheel ready to take over in case of a malfunction. After buying the self-driving truck company Otto in 2016 for a reported $680 million, Uber is now expanding its AI capabilities and plans to test NVIDIA’s latest chips in Otto’s vehicles. It has also partnered with Volvo to create a self-driving fleet of cars and with Toyota to co-create a ride-sharing autonomous vehicle.

Tesla
The first major car manufacturer to come from Silicon Valley, Tesla was also the first to introduce Level 2 autopilot back in 2015. The following year, it announced that all new Teslas would have the hardware for full autonomy, meaning once the software is finished it can be deployed on existing cars with an instant upgrade. Some experts have challenged this approach, arguing that the company has merely added surround cameras to its production cars that aren’t as capable as the laser-based sensing systems that most other carmakers are using.

But the company has collected data from hundreds of thousands of cars, driving millions of miles across all terrains. So, we shouldn’t dismiss the firm’s founder, Elon Musk, when he claims a Level 4 Tesla will drive from LA to New York without any human interference within the first half of 2018.

Winners

Who’s leading the race? Image Credit: IMD

At the moment, the disruptors like Tesla, Waymo, and Uber seem to have the upper hand. While the traditional automakers are focusing on bringing Level 3 and 4 partial automation to market, the new companies are leapfrogging them by moving more directly towards Level 5 full automation. Waymo may have the least experience of dealing with consumers in this sector, but it has already clocked up a huge amount of time testing some of the most advanced technology on public roads.

The incumbent carmakers are also focused on the difficult process of integrating new technology and business models into their existing manufacturing operations by buying up small companies. The challengers, on the other hand, are easily partnering with other big players including manufacturers to get the scale and expertise they need more quickly.

Tesla is building its own manufacturing capability but also collecting vast amounts of critical data that will enable it to more easily upgrade its cars when ready for full automation. In particular, Waymo’s experience, technology capability, and ability to secure solid partnerships puts it at the head of the pack.

This article was originally published on The Conversation. Read the original article.

Image Credit: Waymo Continue reading

Posted in Human Robots

#431872 AI Uses Titan Supercomputer to Create ...

You don’t have to dig too deeply into the archive of dystopian science fiction to uncover the horror that intelligent machines might unleash. The Matrix and The Terminator are probably the most well-known examples of self-replicating, intelligent machines attempting to enslave or destroy humanity in the process of building a brave new digital world.
The prospect of artificially intelligent machines creating other artificially intelligent machines took a big step forward in 2017. However, we’re far from the runaway technological singularity futurists are predicting by mid-century or earlier, let alone murderous cyborgs or AI avatar assassins.
The first big boost this year came from Google. The tech giant announced it was developing automated machine learning (AutoML), writing algorithms that can do some of the heavy lifting by identifying the right neural networks for a specific job. Now researchers at the Department of Energy’s Oak Ridge National Laboratory (ORNL), using the most powerful supercomputer in the US, have developed an AI system that can generate neural networks as good if not better than any developed by a human in less than a day.
It can take months for the brainiest, best-paid data scientists to develop deep learning software, which sends data through a complex web of mathematical algorithms. The system is modeled after the human brain and known as an artificial neural network. Even Google’s AutoML took weeks to design a superior image recognition system, one of the more standard operations for AI systems today.
Computing Power
Of course, Google Brain project engineers only had access to 800 graphic processing units (GPUs), a type of computer hardware that works especially well for deep learning. Nvidia, which pioneered the development of GPUs, is considered the gold standard in today’s AI hardware architecture. Titan, the supercomputer at ORNL, boasts more than 18,000 GPUs.
The ORNL research team’s algorithm, called MENNDL for Multinode Evolutionary Neural Networks for Deep Learning, isn’t designed to create AI systems that cull cute cat photos from the internet. Instead, MENNDL is a tool for testing and training thousands of potential neural networks to work on unique science problems.
That requires a different approach from the Google and Facebook AI platforms of the world, notes Steven Young, a postdoctoral research associate at ORNL who is on the team that designed MENNDL.
“We’ve discovered that those [neural networks] are very often not the optimal network for a lot of our problems, because our data, while it can be thought of as images, is different,” he explains to Singularity Hub. “These images, and the problems, have very different characteristics from object detection.”
AI for Science
One application of the technology involved a particle physics experiment at the Fermi National Accelerator Laboratory. Fermilab researchers are interested in understanding neutrinos, high-energy subatomic particles that rarely interact with normal matter but could be a key to understanding the early formation of the universe. One Fermilab experiment involves taking a sort of “snapshot” of neutrino interactions.
The team wanted the help of an AI system that could analyze and classify Fermilab’s detector data. MENNDL evaluated 500,000 neural networks in 24 hours. Its final solution proved superior to custom models developed by human scientists.
In another case involving a collaboration with St. Jude Children’s Research Hospital in Memphis, MENNDL improved the error rate of a human-designed algorithm for identifying mitochondria inside 3D electron microscopy images of brain tissue by 30 percent.
“We are able to do better than humans in a fraction of the time at designing networks for these sort of very different datasets that we’re interested in,” Young says.
What makes MENNDL particularly adept is its ability to define the best or most optimal hyperparameters—the key variables—to tackle a particular dataset.
“You don’t always need a big, huge deep network. Sometimes you just need a small network with the right hyperparameters,” Young says.
A Virtual Data Scientist
That’s not dissimilar to the approach of a company called H20.ai, a startup out of Silicon Valley that uses open source machine learning platforms to “democratize” AI. It applies machine learning to create business solutions for Fortune 500 companies, including some of the world’s biggest banks and healthcare companies.
“Our software is more [about] pattern detection, let’s say anti-money laundering or fraud detection or which customer is most likely to churn,” Dr. Arno Candel, chief technology officer at H2O.ai, tells Singularity Hub. “And that kind of insight-generating software is what we call AI here.”
The company’s latest product, Driverless AI, promises to deliver the data scientist equivalent of a chessmaster to its customers (the company claims several such grandmasters in its employ and advisory board). In other words, the system can analyze a raw dataset and, like MENNDL, automatically identify what features should be included in the computer model to make the most of the data based on the best “chess moves” of its grandmasters.
“So we’re using those algorithms, but we’re giving them the human insights from those data scientists, and we automate their thinking,” he explains. “So we created a virtual data scientist that is relentless at trying these ideas.”
Inside the Black Box
Not unlike how the human brain reaches a conclusion, it’s not always possible to understand how a machine, despite being designed by humans, reaches its own solutions. The lack of transparency is often referred to as the AI “black box.” Experts like Young say we can learn something about the evolutionary process of machine learning by generating millions of neural networks and seeing what works well and what doesn’t.
“You’re never going to be able to completely explain what happened, but maybe we can better explain it than we currently can today,” Young says.
Transparency is built into the “thought process” of each particular model generated by Driverless AI, according to Candel.
The computer even explains itself to the user in plain English at each decision point. There is also real-time feedback that allows users to prioritize features, or parameters, to see how the changes improve the accuracy of the model. For example, the system may include data from people in the same zip code as it creates a model to describe customer turnover.
“That’s one of the advantages of our automatic feature engineering: it’s basically mimicking human thinking,” Candel says. “It’s not just neural nets that magically come up with some kind of number, but we’re trying to make it statistically significant.”
Moving Forward
Much digital ink has been spilled over the dearth of skilled data scientists, so automating certain design aspects for developing artificial neural networks makes sense. Experts agree that automation alone won’t solve that particular problem. However, it will free computer scientists to tackle more difficult issues, such as parsing the inherent biases that exist within the data used by machine learning today.
“I think the world has an opportunity to focus more on the meaning of things and not on the laborious tasks of just fitting a model and finding the best features to make that model,” Candel notes. “By automating, we are pushing the burden back for the data scientists to actually do something more meaningful, which is think about the problem and see how you can address it differently to make an even bigger impact.”
The team at ORNL expects it can also make bigger impacts beginning next year when the lab’s next supercomputer, Summit, comes online. While Summit will boast only 4,600 nodes, it will sport the latest and greatest GPU technology from Nvidia and CPUs from IBM. That means it will deliver more than five times the computational performance of Titan, the world’s fifth-most powerful supercomputer today.
“We’ll be able to look at much larger problems on Summit than we were able to with Titan and hopefully get to a solution much faster,” Young says.
It’s all in a day’s work.
Image Credit: Gennady Danilkin / Shutterstock.com Continue reading

Posted in Human Robots

#431828 This Self-Driving AI Is Learning to ...

I don’t have to open the doors of AImotive’s white 2015 Prius to see that it’s not your average car. This particular Prius has been christened El Capitan, the name written below the rear doors, and two small cameras are mounted on top of the car. Bundles of wire snake out from them, as well as from the two additional cameras on the car’s hood and trunk.
Inside is where things really get interesting, though. The trunk holds a computer the size of a microwave, and a large monitor covers the passenger glove compartment and dashboard. The center console has three switches labeled “Allowed,” “Error,” and “Active.”
Budapest-based AImotive is working to provide scalable self-driving technology alongside big players like Waymo and Uber in the autonomous vehicle world. On a highway test ride with CEO Laszlo Kishonti near the company’s office in Mountain View, California, I got a glimpse of just how complex that world is.
Camera-Based Feedback System
AImotive’s approach to autonomous driving is a little different from that of some of the best-known systems. For starters, they’re using cameras, not lidar, as primary sensors. “The traffic system is visual and the cost of cameras is low,” Kishonti said. “A lidar can recognize when there are people near the car, but a camera can differentiate between, say, an elderly person and a child. Lidar’s resolution isn’t high enough to recognize the subtle differences of urban driving.”
Image Credit: AImotive
The company’s aiDrive software uses data from the camera sensors to feed information to its algorithms for hierarchical decision-making, grouped under four concurrent activities: recognition, location, motion, and control.
Kishonti pointed out that lidar has already gotten more cost-efficient, and will only continue to do so.
“Ten years ago, lidar was best because there wasn’t enough processing power to do all the calculations by AI. But the cost of running AI is decreasing,” he said. “In our approach, computer vision and AI processing are key, and for safety, we’ll have fallback sensors like radar or lidar.”
aiDrive currently runs on Nvidia chips, which Kishonti noted were originally designed for graphics, and are not terribly efficient given how power-hungry they are. “We’re planning to substitute lower-cost, lower-energy chips in the next six months,” he said.
Testing in Virtual Reality
Waymo recently announced its fleet has now driven four million miles autonomously. That’s a lot of miles, and hard to compete with. But AImotive isn’t trying to compete, at least not by logging more real-life test miles. Instead, the company is doing 90 percent of its testing in virtual reality. “This is what truly differentiates us from competitors,” Kishonti said.
He outlined the three main benefits of VR testing: it can simulate scenarios too dangerous for the real world (such as hitting something), too costly (not every company has Waymo’s funds to run hundreds of cars on real roads), or too time-consuming (like waiting for rain, snow, or other weather conditions to occur naturally and repeatedly).
“Real-world traffic testing is very skewed towards the boring miles,” he said. “What we want to do is test all the cases that are hard to solve.”
On a screen that looked not unlike multiple games of Mario Kart, he showed me the simulator. Cartoon cars cruised down winding streets, outfitted with all the real-world surroundings: people, trees, signs, other cars. As I watched, a furry kangaroo suddenly hopped across one screen. “Volvo had an issue in Australia,” Kishonti explained. “A kangaroo’s movement is different than other animals since it hops instead of running.” Talk about cases that are hard to solve.
AImotive is currently testing around 1,000 simulated scenarios every night, with a steadily-rising curve of successful tests. These scenarios are broken down into features, and the car’s behavior around those features fed into a neural network. As the algorithms learn more features, the level of complexity the vehicles can handle goes up.
On the Road
After Kishonti and his colleagues filled me in on the details of their product, it was time to test it out. A safety driver sat in the driver’s seat, a computer operator in the passenger seat, and Kishonti and I in back. The driver maintained full control of the car until we merged onto the highway. Then he flicked the “Allowed” switch, his copilot pressed the “Active” switch, and he took his hands off the wheel.
What happened next, you ask?
A few things. El Capitan was going exactly the speed limit—65 miles per hour—which meant all the other cars were passing us. When a car merged in front of us or cut us off, El Cap braked accordingly (if a little abruptly). The monitor displayed the feed from each of the car’s cameras, plus multiple data fields and a simulation where a blue line marked the center of the lane, measured by the cameras tracking the lane markings on either side.
I noticed El Cap wobbling out of our lane a bit, but it wasn’t until two things happened in a row that I felt a little nervous: first we went under a bridge, then a truck pulled up next to us, both bridge and truck casting a complete shadow over our car. At that point El Cap lost it, and we swerved haphazardly to the right, narrowly missing the truck’s rear wheels. The safety driver grabbed the steering wheel and took back control of the car.
What happened, Kishonti explained, was that the shadows made it hard for the car’s cameras to see the lane markings. This was a new scenario the algorithm hadn’t previously encountered. If we’d only gone under a bridge or only been next to the truck for a second, El Cap may not have had so much trouble, but the two events happening in a row really threw the car for a loop—almost literally.
“This is a new scenario we’ll add to our testing,” Kishonti said. He added that another way for the algorithm to handle this type of scenario, rather than basing its speed and positioning on the lane markings, is to mimic nearby cars. “The human eye would see that other cars are still moving at the same speed, even if it can’t see details of the road,” he said.
After another brief—and thankfully uneventful—hands-off cruise down the highway, the safety driver took over, exited the highway, and drove us back to the office.
Driving into the Future
I climbed out of the car feeling amazed not only that self-driving cars are possible, but that driving is possible at all. I squint when driving into a tunnel, swerve to avoid hitting a stray squirrel, and brake gradually at stop signs—all without consciously thinking to do so. On top of learning to steer, brake, and accelerate, self-driving software has to incorporate our brains’ and bodies’ unconscious (but crucial) reactions, like our pupils dilating to let in more light so we can see in a tunnel.
Despite all the progress of machine learning, artificial intelligence, and computing power, I have a wholly renewed appreciation for the thing that’s been in charge of driving up till now: the human brain.
Kishonti seemed to feel similarly. “I don’t think autonomous vehicles in the near future will be better than the best drivers,” he said. “But they’ll be better than the average driver. What we want to achieve is safe, good-quality driving for everyone, with scalability.”
AImotive is currently working with American tech firms and with car and truck manufacturers in Europe, China, and Japan.
Image Credit: Alex Oakenman / Shutterstock.com Continue reading

Posted in Human Robots