Your Next Holiday Party, Catering By Bot

It is that time of year, when regardless of your religion your are invited to a million Holiday parties.  While many go to socialize, network, and toast the past year’s successes, few think about the process that brought that chicken finger into existence.  As urbanites we are far very removed from the disgusting manufacturing of meat,  just watch two minutes of Fast Food Nation and you may consider becoming vegetarian…

The good news is robots might be the best worker to shield us from the ugliness of food production (ignorance is blissfully tasteful). This week in Norway, a team of engineers from SINTEF research group announced a new robotic way of pulling chicken breasts off the bird using their robot – Gribbot.  “Gribb” is Norwegian for vulture, and the machine is said to resemble a vulture’s beak (below).


Pulling chicken breasts off the bone can be a difficult process, and often results in flesh being wasted by getting left behind. In a factory setting, that means slower processing times, and less meat to sell.  Chickens are big business. In the United States alone 8.7 billion chickens are processed every year. That’s 36 billion pounds (16 billion kg) of poultry of which the average American consumes 50 pounds (23 kg). On such a scale, costs due to inefficient processing and the need for skilled labor adds up very quickly. Over the years, the meat industry have managed to automate many tasks so that, for example, making bacon is now a largely hands-off affair, but chickens are very irregularly and individually shaped, so automation has passed them by, until now.

Gribbot utilizes a Kinect camera for 3D vision, along with a flexible-fingered grasping hand, and an algorithm that guides that hand’s movements based on what the camera sees. Challenges lie in the fact that the meat is slippery and hard to hold onto, plus reflections on its shiny surface can make 3D imaging difficult.  Nevertheless, Gribbot is apparently quite adept at grasping chicken carcasses, identifying the breast meat, and then simultaneously scraping and pulling it off the bone. It is hoped that once perfected, the robot “will make Norwegian food production more sustainable, both in terms of profitability and utilization of raw materials.”   SINTEF has a lot of experience their previous invention was a fish-filleting robot below:


Back here at home, Georgia Tech Research Institute (GTRI) have created the ultimate bird deboning weapon.  Using 3D imaging technology, the “Intelligent Cutting and Deboning System” developed at GTRI can debone an entire chicken with the skill of a human butcher and has the potential of saving the poultry industry millions of dollars by reducing costs and waste.

“Each bird is unique in its size and shape, so we have developed the sensing and actuation needed to allow an automated deboning system to adapt to the individual bird, as opposed to forcing the bird to conform to the machine,” says Gary McMurray, chief of GTRI’s Food Processing Technology Division.

This is really ground breaking as the robot visually seeks out fixed landmarks on the chicken and uses these as part of an algorithm that estimates the internal structure of the chicken and plans the cuts. These cuts are designed to be precise with the intention of maximizing the amount of meat removed and minimizing the number of bone fragments. This is achieved by the robot’s two arms. One is a cutting arm with two degrees of movement and the other is a holding arm with six degrees of movement. In concert, these arms allow the robot to position the chicken and make the required cuts.


But this wouldn’t be of much help without the final component. The robot can’t see through the chicken while cutting, so it relies on a sort of mechanical “touch” to see how things are going. The arms include a forced-feedback system that tells the robot when it is meeting resistance, such as a bone or tendon. If it meets a hard resistance, it avoids this as a bone. If the resistance only slows the knife, then it’s a tendon or ligament. With these two cues, the robot can stay close to the bone without chipping it and by knowing when it cuts a tendon or ligament, it can remove the meat cleanly and completely, though this is still something of an art.

“Fine tuning is needed to adjust the force thresholds to be able to tell the difference between meat, tendon, ligaments and bone – each of which have different material properties,” said research engineer Ai-Ping Hu (seen above).

The chicken-deboning robot is still undergoing testing and development by GTRI under funding by the state of Georgia through the Agricultural Technology Research Program, but one day soon these robots will be able to replace the abused workers who are longing for better jobs upstream.  More meat, less killing, makes chicken little worry free (well, sort-of).

Robotic Elves Get To Work

While Jeff Bezos has yet to grow a white beard, or gain 300 lbs, Amazon is the closest we will get to Santa’s workshop.  The Northpole might not be as functional as the 80 worldwide fulfillment centers that this merry-man has under his belt.  The 100,000 plus employees are being helped this season with a robotic elf army, and this is welcomed news as Cyber Monday’s sales were up 15% from last year.

As we reported two years ago, Amazon acquired Kiva Systems for $775 million. Kiva, a Massachusetts-based startup, makes warehouse robots and software to automate the most mundane parts of filling online orders, e.g. sorting, pick-n-pack, and moving inventory. This past Sunday, on the eve of the high point of the internet retail year, Amazon opened its doors to show off its army of Kiva droids that are taking its warehouses to the next level of consumer gratification.  The big surprise is that Kiva is now joined by “Robo-Stow,” one of the largest robotic arm in existence, and various vision AI systems. However, I’m sorry to report, there were no drones flying above delivering packages like Rudolph and his reindeer buddies.

In short, Amazon said it’s expanding the use of “Robo-Stow,” which it describes as “one of Earth’s largest robotic arms moving large quantities of inventory for customer order fulfillment.” It’s also using new “vision systems” to unload delivery trucks in a half-hour, as opposed to the hours it took previously.

“The Amazon fulfillment teams are dedicated to innovating in our fulfillment centers to increase speed of delivery while enabling greater local selection at lower costs for our customers,” said Dave Clark, Amazon’s senior vice president of worldwide operations and customer service.


The first impression at any Amazon warehouse—the company calls them “fulfillment centers”—is the noise. Plastic bins rattling over rollers, conveyer belts brimming with packages, hissing machines slapping labels onto boxes—the warehouse is in its self a well oiled machine in constant motion. Unlike older versions of its fulfillment centers, this one has zones of quiet at its heart, and this is the den of the Kiva bots.

Inside black chain-link cages on four floors, swarms of Kiva bots glide silently along an invisible grid. The squat orange automatons look like the smaller cousins of bumper cars, except they never collide. On their “backs,” the 320-pound robots carry shelves standing several feet taller than their human attendants outside the cage and weighing up to 750 pounds . Each shelf is stuffed with bins on all four sides that hold inventory—coffee mugs, Crayola markers, GoPro lens protectors, USB cables, or just about any other of around 3.5 million different items stocked by Amazon at its San Francisco warehouse, known by its call sign as OAK4.

The moving shelves serve two main purposes. When inventory arrives at the fulfillment center, items have to wait somewhere before someone orders them. In the traditional process known as “stowing,” a person walks the aisles with a cart and shelves the items. With the Kiva robots, Amazon workers wait at stations for the shelves to come to them, where they load up each one with inventory while the other robots queue politely behind. In the second part of the process, called “picking,” orders come in, and pre-Kiva, pickers would walk the aisles to the bins storing the needed items. Now the shelves come to the pickers instead.

Amazon gains two major advantages by using robots instead of people to move the merchandise, says Dave Clark. Because aisles aren’t needed between shelves to make room for human pickers, more items can be stored more closely together, he says, enabling a fulfillment center to hold more stuff. And the time saved by eliminating the need for humans to walk to where items are makes the movement of items in and out the door more efficient.

“Kiva’s doing the part that’s not that complicated. It’s just moving inventory around,” Clark says. “The person is doing the complicated work, which is reaching in, identifying the right product, making sure it’s the right quality, making sure it’s good enough to be a holiday gift for somebody.”

Robots queue at an Amazon "stow station."

Right now, Clark says OAK4 holds 21 million items total in its inventory. Once the fulfillment center is entirely built out with its full Kiva cohort, it will hold 26 million items total, representing 5 million different products. On a peak day, he says, OAK4 can ship out 700,000 items. By the time it’s fully operational, he says, that number will rise as high as 1.5 million.

In the US, Amazon now has a workforce of 15,000 Kiva robots deployed across 10 of its 50 domestic fulfillment centers. Another fulfillment center in California has the massive robot arm, the “Robo-Stow,” capable of lifting entire pallets from one floor to another. The efficiency gains mean moving more orders more quickly, which among other effects will eventually push back the current noon cutoff time for same-day delivery orders to later in the afternoon.


Since displaying its robotic muscle, many reporters have innocently asked, “But if robots move Amazon’s merchandise faster while costing less overall, won’t Amazon try to find ways to replace as many of those human workers as possible?”  Clark responded correctly that the rise in productivity will give Amazon the means to grow. And growth means Amazon will need to hire more people.

“As you’re getting efficient in one thing as it matures, you’re investing in something new in a new place,” Clark says. “I see that cycle continuing for a long time.”

While some were quieted, others still criticize with the same fingers that just ordered their latest Amazon Prime delivery.  The holiday season is upon us, and more important than presents is giving.  Here is one suggestion of a worthy charity: UJA NY.

Solar Robots Bring Energy Efficiency

Last week, I met with Jim Tisch of Loews Corporation.  A lot of our conversation centered around the dramatic drop in gas prices.  OPEC is clearly manipulating the US oil production by flooding the market with cheap Arabian barrels, and thus slowing down domestic production.  If cartels can use their weight to manipulate  our markets, it doesn’t matter how great our technology is it will always be subject to outside factors.  The words of the Lorax ring perfectly clear, “UNLESS,” — Unless we achieve energy independence through alternative methods like solar.

Solar fields are growing faster than corn, but the big problem is dust, which reduces efficiency by over 35%. This brings us to the robots designed by Israeli startup Ecoppia, as an alternative to  human workers to hose and wipe down panels manually:


Dirty panels produce less electricity, but the need to use water for cleaning those panels, especially in dry regions, makes even a clean power project less eco-friendly. And in certain remote corners, water extracted from the ground is too brackish for use without being treated, which adds to the production cost of a solar power plant.  In dusty areas such as the Middle East and India, solar panels could lose electricity production by 10 percent to 35 percent over time if they remain unwashed, Eran Meller, CEO of Ecoppia, was recently interview by fellow blogger, GIGAOM.

Ecoppia’s robots dry clean each panel and move from the top to the bottom of a row of panels. The Israeli startup found a loyal customer in Arava Power, with which Ecoppia installed the first set of its robots on a solar farms (5 MW total) earlier this year in the Negev desert. Ecoppia is installing more robots in other Arava projects.


“It doesn’t pay to manually clean thousands of panels in hundreds of acres of arid desert fields,” said Jon Cohen, Arava’s CEO. “Now we have a process that costs less, and above that we are upping the output.” Using the robots so far has led to about 2-3 percent more electricity production than employing humans, Cohen said.

The challenge of keeping solar panels dust free will grow as more solar power projects are built worldwide. In many cases, cheap labor and ample water supply will continue to make manual washing the low-cost choice for solar power plant owners. But for companies with projects in different climates — and the need to show they run a low-carbon, sustainable operation to secure permits or dodge lawsuits from environmental groups — a less energy intensive cleaning process could be desirable.

SunPower, which builds solar power projects around the world, bought Greenbotics a year ago after trying out the California startup’s technology in a solar farm it built in the state’s Central Valley.  The big selling point of Greenbotics is that its technology uses up to 90 percent less water than manual cleaning.

Ecoppia was founded in January 2013 but started its development work a few quarters before that. The company has raised an undisclosed amount from the Swarth Group, GlenRock and Gandyr.

Each robot, which weighs about 86 kilograms (190 pounds), is essentially two large microfiber brushes on eight wheels, with the brushes rotating at a high speed to generate airflow as they move down the panel. The airflow removes a bulk of the dust while the brushes get rid of the rest. The robot runs on two 12-volt lead-acid batteries at night. Solar electricity recharges the batteries during the day. After the robot completes its task, it returns to a docking station and uses the rotational energy to get rid of the dust captured by the microfiber.

Ecoppia has designed its robotic system to perform optimally in a row of solar panels that runs 300 meters by 6 meters (984 feet by 20 feet), Meller said. Each robot can take care of hundreds of panels each night, depending on the size and configuration of the installation. Power plant operators can control the robots remotely and receive data about the machines’ performance and maintenance needs.

Ecoppia makes money by selling and installing the robots and providing maintenance and data analytics. The cost to hire Ecoppia to engineer and install its robots runs from $0.03 to $0.06 per watt, Meller said.

With about one year of field data of its robots’ performance, the startup projects that its equipment and services could save 840 million liters of water for a 300 MW solar park over 20 years while increasing electricity sales by $180 million, Meller said. Of course, those projected savings and revenues will vary widely in different countries or even within a country, depending on the local operational costs and how much the utilities are willing to pay for power. Arava, for example, is cleaning its solar panels nightly in Israel while in California, SunPower is cleaning its panels several times a year.

Currently, Ecoppia’s robots are cleaning about 500,000 panels per month. It will be installing robots for other Arava projects, including 40 megawatts that have yet to be completed, that will bring the monthly total to 10 million by the end of 2015, Meller said. The robotics developers are working on entering the U.S. market next year.

Now all we need to do is bring E4 robotics to mars to be able to clean and activate dormant (dust-covered) solar-powered rovers…

RoboCop Patrols Microsoft’s Parking Lot

Living in New York, everyone is focused on crime.  We event have a data bank of daily crime called, COMPSTAT.   However, recent waves of bad policing has made cops embrace new technologies to protect themselves from public scrutiny.  The question is not if will robots replace meter maids, traffic cops and other tertiary policing, but when…

As an example, earlier this week Microsoft hired high-tech security guards to parole the streets, protecting people in California’s Silicon Valley. The Knightscope K5 robots are fighting crime with lasers, GPS and heat-detecting technology. The R2-D2 look alikes patrol autonomously in the set perimeter and record activity around its path.

Providing a “commanding but friendly presence,” the robots were hired to intimidate potential criminals. The new high-tech security guards have the ability to catch a criminal red-handed. “The robot is looking at the video, listening for glass breakage, any loud sound that breaking in would cause,” Knightscope co-founder Stacy Stephens says. “We’ll get the license plate, picture of the vehicle, geotag location, and time.”

The robots stand five-feet-tall and weigh 300 pounds. The robots have laser scanners, a thermal imaging system and the ability to read 300 car license plates in a minute. Its 360-degree HD  surveillance camera streams live video to a command center. And yes, the robots can also detect odors.

But these real-life R2-D2s are not equipped with weapons, and can only contact human security officers to the scene after sounding off an alarm when something goes wrong.  The robots last an entire day with just one charge. When the robot notices its battery is low, it will plus itself into a charger for 20 minutes until it is fully charged. Silicon Valley’s computer giant Microsoft have already stationed four robot security guards on its campus. Ideal for college campuses, malls, or other busy outdoors areas, Knightscope says that approximately four dozen companies remain on a waiting list for the K5 robots.

K5 is not unique but part of new wave of robotic recruits for the policing of the future, and you thought it was just a movie.

Robot Boldly Goes Where No Thing Has Gone Before…

The biggest news to hit the robotic sphere this week are Rosetta & Philae the European robot comet voyagers.  This dynamic duo which has been led by the ESA, the EU’s equivalent to NASA, is a tag team effort to better understand the Universe through attaching scientific instruments to moving comets.   Below are some of the first videos and images taken by the robots (note: Rosetta hovers above while Philae attaches to the comet ice rock below).

 The first picture below is from the lander separating from its “mothership”, Rosetta.  The mission success will determine on Philae’s ability to analyze the icy rock of the comet to better understand the formation of our Solar System.  Most scientist believe that comets contain the oldest material known in our Solar System.

First image
Success! Philae landed just missing the spiral above that would have destroyed the spacecraft:
In addition to the spiky rock particles, challenges to the mission include a very low gravity on the 4km-wide ice mountain. Philae could simply bounce back to space, as its foot screws and harpoons that fasten it into position failed to deploy.  Mission command said they might use its exploration drill as a back up.

During the mission, which started yesterday afternoon, Philae will take a pictures of its surroundings like the one above and collect data on early water particles.  Figures crossed as its outcome is highly uncertain.  Rather than breaking down the parts composed in these voyagers, we have posted below the schematics that can be enlarged by clicking on the image.  This is one small step for Philae, one giant step for robotkind.