An HSBC executive has said that the bank’s blockchain-based system has helped it cut the costs of settling foreign exchange trades.
Speaking to Reuters, Mark Williamson, chief operating officer of FX cash trading and risk management, who oversees the blockchain project, said that its HSBC FX Everywhere platform saved it 25 percent as compared with traditional methods.
Last month, the bank announced it had settled more than $250 billion in transactions using its HSBC FX Everywhere platform.
It said then that it had settled 3 million foreign exchange transactions and made a further 150,000 payments over the digital ledger system, which it has been using over the last year “to orchestrate payments across HSBC’s internal balance sheets.”
In the Reuters report, Williamson said that HSBC processes from 3,500 to 5,000 trades a day on FX Everywhere, with trades now being worth $350 billion.
“We’re able to demonstrate that this is not a one-off proof of concept or just one or two trades,” Williamson said.
Reuters further quoted him as saying that a “significant” amount of internal money flows are likely to be settled on the DLT system.
HSBC has been experimenting wit blockchain for some time. Since joining blockchain consortium startup R3 in 2015, it has teamed up with Bank of America and the Singapore government on a blockchain supply chain trial
It’s also joined work on the Utility Settlement Coin (USC) project, designed to make it easier for global banks to conduct a variety of transactions with each other using collateralized assets on a custom-built blockchain.
A senior executive at Western Union has announced the company is getting “ready” for a surge in cryptocurrencies, and is currently trialing some of the technology developed by Ripple Labs for use in its settlements services.
Molly Shea, the General Manager of Asia Pacific at Western Union, explained in the Analyse Asia podcast that the financial services company – which reported revenue of $1.4bn in Q4 2018 – was preparing itself for a future where digital assets begin to “take off”.
“When you think about cryptocurrencies, if those start to take off, we need to be ready,” Shea said. “We need to be ready form a technology perspective; regulation’s got to be there, but you constantly have to be looking for those trends where the customers are looking…and make sure that you’re ready to meet those needs and expectations. “
As part of this preparation, Shea said, Western Union is exploring legitimate and widely acceptable use cases for cryptocurrencies. One such area is in settlements; the company is currently performing“settlement tests” on whether Ripple could be used in “certain corridors,” particularly with the US dollar and Mexican peso.
Crypto Briefing reached out to Western Union for additional information and comment on its Ripple trials, but did not receive a response by press time.
Western’s Union With Ripple
Established in 1851, 168 years ago, Western Union provides remittance services, allowing customers to send funds across borders and in multiple currencies. The company sent more than $300bn in 2018; its network has more than 550,000 retail locations; more than all the Starbucks and McDonald’s outlets in the world, Shea claims.
Although the XRP cryptocurrency – part of the xRapid solution – is perhaps the most famous product offered by Ripple Labs, it isn’t the only one. The company also offers an xCurrent solution, which uses blockchain technology, but not necessarily XRP, for banks and other financial institutions to communicate with one another before, during and after a cross-border transaction.
A source at the National Bank of Kuwait told Crypto Briefing in October last year that the bank was waiting for regulatory approval to use xCurrent. The service finally went live at the end of December.
Collaboration between Ripple Labs and Western Union has been known for some time; the company began testing Ripple’s blockchain towards the end of 2017. In 2016, Western Union invested in Digital Currency Group (DCG), a venture capital company, which has also made investments into Ripple.
Although CEO Hikmet Ersek admitted last summer that the trials had yet to yield any cost savings, President Odilon Almeida told Reuters in December that there was little difference between crypto and digital fiat payments.
Adding support would not likely represent much of a challenge.
For the first time, physicists have built reliable, efficient graphene nanomachines that can be fabricated on silicon chips. They could lead to even greater miniaturization.
The chances are that you own a microelectromechanical device—probably dozens of them. These devices fill the modern world. They make possible the accelerometers in smartphones, the microphones in laptops, and the micromirrors in digital projectors, to name just a few.
They are typically a few micrometers in size, tiny by any standards. But scientists and engineers want them even smaller—on the nanometer scale, if possible. At that size, these machines can work as simple switches in logic and memory devices, raising the prospect of more powerful and more efficient data-processing devices.
These micromachines are generally carved into silicon chips. But as they get smaller, silicon switches become less efficient because they leak current when they are off. A better option is a graphene switch, which is easy to carve on a nanometer scale and relatively straightforward to build into conventional silicon chips. Neither does it leak current when it is off.
But there is a problem. When graphene touches silicon, it tends to stick fast. Imagine a switch consisting of a flexible graphene bar that forms a circuit when the bar touches a silicon electrode. If the bar sticks to the electrode, it cannot be switched off again.
This problem is known as stiction. And despite significant financial investment in graphene research by governments all over the world, nobody has found a good way to solve it.
Enter Kulothungan Jothiramalingam at the Japan Advanced Institute of Science and Technology and colleagues, who have found a solution. Using it, they have created graphene-based nanoelectromechanical devices that can act as switches and even as logic gates.
Their method is straightforward. They coat a silicon chip with nanocrystalline graphene, which sticks fast to the surface. Then they cover this with a layer of hydrogen silsesquioxane, which acts as a resist and can be carved into various shapes. On top of this they place another layer of graphene.
The trick is to carve the top layer of graphene into a bar shape that is anchored at both ends by electrodes. Then they remove the hydrogen silsesquioxane layer underneath part of the graphene bar to leave it suspended above the graphene layer.
Bending this bar is simple. A potential difference between the layers creates a force that bends the bar to toward the chip. When it touches this lower surface, it forms a circuit, a process that can be exploited for logic and for data storage.
That’s the switch. And because the two surfaces that come into contact are both graphene, there is no stiction. Switching off the potential difference releases the bar, which springs back into its original position.
Jothiramalingam and co used this approach to build a variety of proof-of-principle nano-switches, including single switches and an array. They say the devices work well with low voltages of just 1.5 volts and that in the off state, there is very little current leakage because the graphene bars are well insulated from other conducting layers.
There are some challenges, though. For example, the shape and size of the graphene beam and its distance from the lower layer need to be optimized to achieve reliable switching. But this should be a straightforward engineering problem.
Once that is solved, more complex devices become possible. The team has designed a range of more complex switches including an AND logic gate and a three-terminal switch in which they place three layers of graphene on top of each other, separated by an insulating layer of hydrogen silsesquioxane.
That’s interesting work with the potential to make nanoelectromechanical devices even smaller, based on the promise of the wonder material that is graphene.
Ref: arxiv.org/abs/1901.07754 : Stacking of Nanocrystalline Graphene for Nano-ElectroMechanical (NEM) Actuator Applications
Scientists have used a Nobel-prize winning Chemistry technique on a mixture of metals to potentially reduce the cost of fuel cells used in electric cars and reduce harmful emissions from conventional vehicles.
The researchers have translated a biological technique, which won the 2017 Nobel Chemistry Prize, to reveal atomic scale chemistry in metal nanoparticles. These materials are one of the most effective catalysts for energy converting systems such as fuel cells. It is the first time this technique has been used for this kind of research.
The particles have a complex star-shaped geometry and this new work shows that the edges and corners can have different chemistries which can now be tuned to reduce the cost of batteries and catalytic convertors.
The 2017 Nobel Prize in Chemistry was awarded to Joachim Frank, Richard Henderson and Jacques Dubochet for their role in pioneering the technique of ‘single particle reconstruction’. This electron microscopy technique has revealed the structures of a huge number of viruses and proteins but is not usually used for metals.
Now, a team at the University of Manchester, in collaboration with researchers at the University of Oxford and Macquarie University, have built upon the Nobel Prize winning technique to produce three dimensional elemental maps of metallic nanoparticles consisting of just a few thousand atoms.
Published in the journal Nano Letters, their research demonstrates that it is possible to map different elements at the nanometre scale in three dimensions, circumventing damage to the particles being studied.
Metal nanoparticles are the primary component in many catalysts, such as those used to convert toxic gases in car exhausts. Their effectiveness is highly dependent on their structure and chemistry, but because of their incredibly small structure, electron microscopes are required in order to provide image them. However, most imaging is limited to 2D projections.
“We have been investigating the use of tomography in the electron microscope to map elemental distributions in three dimensions for some time,” said Professor Sarah Haigh, from the School of Materials, University of Manchester. “We usually rotate the particle and take images from all directions, like a CT scan in a hospital, but these particles were damaging too quickly to enable a 3D image to be built up. Biologists use a different approach for 3D imaging and we decided to explore whether this could be used together with spectroscopic techniques to map the different elements inside the nanoparticles.”
“Like ‘single particle reconstruction’ the technique works by imaging many particles and assuming that they are all identical in structure, but arranged at different orientations relative to the electron beam. The images are then fed in to a computer algorithm which outputs a three dimensional reconstruction.”
In the present study the new 3D chemical imaging method has been used to investigate platinum-nickel (Pt-Ni) metal nanoparticles.
Lead author, Yi-Chi Wang, also from the School of Materials, added: “Platinum based nanoparticles are one of the most effective and widely used catalytic materials in applications such as fuel cells and batteries. Our new insights about the 3D local chemical distribution could help researchers to design better catalysts that are low-cost and high-efficiency.”
“We are aiming to automate our 3D chemical reconstruction workflow in the future”, added author Dr Thomas Slater.“We hope it can provide a fast and reliable method of imaging nanoparticle populations which is urgently needed to speed up optimisation of nanoparticle synthesis for wide ranging applications including biomedical sensing, light emitting diodes, and solar cells.”
Overstock’s land registration subsidiary will be working with the municipality of Tulum in Quintana Roo in Mexico to develop a land registration platform, as reported by Nasdaq, February 4, 2019.
Medici in Mexico
Blockchain has previously been used for land registration in the United Kingdom by the government, and now the same application will be used in Mexico after a Memorandum of Understanding (MoU) was signed between the Mexican government and Overstock’s Land registration division on February 4, 2019.
As a result of this agreement, a digital platform will be developed to collect land registration data and issue certificates to public landowners. This platform will be developed as a partnership between Medici Land Governance (MLG) and the municipality of Tulum in Quintana Roo. With time, the platform will also be involved in the automatic storage of public land transactions such as transfer of ownership.
Easing the Process
This new partnership will help in the organization of the land registration process saving time and energy and preventing disputes.
Víctor Mas Tah, mayor of the municipality of Tulum, Quintana Roo, Mexico said:
“The signing of this Memorandum of Understanding between the Municipality of Tulum and Medici Land Governance represents the beginning of a new territorial ordering stage for the digitalization of land ownership and related processes”
Land ownership is one of the most complex parts of public life as oftentimes, transfers of ownership are not properly documented and as such, it is not uncommon to see disputes over land ownership last months and even years. This is why land registration has been so heavily embraced by world governments such as New South Wales and even Zambia who signed a similar memorandum of understanding with MLG.
At the time, a press release wrote:
“Without formal ownership, individuals struggle to obtain access to credit and public services, while governments are limited in their ability to collect taxes, enforce property rights, and plan for economic expansion and innovation”
There is also a benefit to the governments themselves; by having a secure land registration process, they have access to better international opportunities and can easily attract investors. This belief has been backed up by Medici Land Governance CEO, Ali El Husseini, who said:
“Mexico’s adoption of advanced technology in their land registry will increase opportunities for individuals to strengthen their connections to the global economy through rightful ownership of land.”
Panasonic wasted no time lowering its guidance after its bread-and-butter customer, Tesla, announced it was buying another battery supplier to power its electric vehicles.
The lucky company Tesla chose to replace Japan-based Panasonic is California-based Maxwell Technologies.
On the news, Panasonic lowered its profit expectations by 9%. The possible loss of Tesla isn’t the only culprit that led to the lowered guidance. The struggling tech player revealed it was also being hurt by weak demand in China for auto components and factory equipment. China’s slowing economy and the overhang of trade wars have weighed on countries and tech companies all over the world.
Musk Praised Panasonic Just Three Months Ago
Back in November, it appeared that the partnership between Tesla and Panasonic was going well. CEO Elon Musk took to Twitter to sing the praises about Panasonic helping it boost profits.
However, Musk had other plans. CCN raised the caution flag on Panasonic last month. We pointed to Tesla’s November indication that it would diversify its sources after experiencing several problems with its Model 3 supply chain.
On the heels of that announcement, rumors swirled that Tesla was on the lookout for a new battery supplier.
In previous reports, CCN noted that Panasonic was also feeling the effects of the possibility of losing Tesla. Its stock price was down by more than 2% on the news that the carmaker was looking for a new supplier.
Tesla Giveth Then Taketh Away
Interestingly, when this supply agreement was announced, Tesla stated:
“The agreement supplies Tesla with Panasonic’s lithium-ion battery cells to build more than 80,000 vehicles over the next four years. It guarantees the availability of enough cells in 2012 to meet Tesla’s aggressive production ramp-up and fulfillment of more than 6,000 existing Model S reservations. This supply agreement helps ensure Tesla will meet its cost and margin targets for Model S.”
The purchase of Maxwell Technologies comes less than a decade since Musk and company inked the deal with Panasonic. The electric vehicle maker had lauded Panasonic as being a battery cell manufacturer and a diverse supplier to the global automotive industry.
Musk’s Always Up To Something, Could Be Good This Time
The move is a disappointing one for Panasonic, but it’s a solid one for Tesla, which has been under financial pressure.
In January, CCN reported that Tesla enjoyed a solid Q4 2018 with record production and delivery numbers driving the company’s first profit in two years. The company posted a net profit of $311.5 million and $891 million in free cash flow. However, the company’s stock price slumped 9% after it failed to meet investor targets for delivery and production numbers.
Owning battery supplier Maxwell should help the company lower its operating costs. The heavily indebted electric car company, whose CFO stepped down just four days ago, is making the acquisition in an all-stock deal.
A Tesla stock shorter (Musk has extreme disdain for them) responded to the Maxwell announcement with this tweet.
1/ ok $TSLAQ, we've had a couple hours to digest the news of the $TSLA acquisition of $MXWL. Like with everything Tesla, it looks funny at first, second, and third glance. Here's what I've gathered as possible reasons/considerations behind the acquisiton. https://t.co/IzzynMr7rK
Elon Musk posted four tweets last night demonstrating the new Raptor rocket engine. SpaceX engineers have been hard at work finalizing the next-generation rocket technology that will eventually power trips for the exploration of Mars.
While most of America was focused on the Super Bowl, Musk was in Texas at the SpaceX testing site. The rockets had recently been shipped from California to the testing site near McGregor, Texas. Musk and his team were up late preparing for the first test late Sunday night. A few hours later, the first video went live on Twitter.
Intended to be reusable, the engines use cryogenic liquid methane and liquid oxygen. Previous SpaceX rockets used RP-1 Kerosene and liquid oxygen. Developing the rockets in-house was part of Musk’s business plan from the start, according to an early SpaceX investor.
When the rockets were first shipped from California to Texas, Musk noted that the company is working hard to get a moon-worthy rocket to ready. The ultimate goal of the Raptor rockets is still for Mars exploration. The Raptor is intended to replace existing rockets already in production, the Falcon 9, Falcon Heavy, and Dragon.
The Raptor product is currently on track. Musk has previously tweeted that the company will attempt a moon trip first.
While SpaceX keeps its eyes on human exploration of Mars, heavily developing the Raptor, it is under financial pressure to actually develop revenues. The company is good at winning open contracts from governments and large companies around the world, but recently suffered a spate of layoffs. Musk blamed these layoffs on the “absolutely insane” Starlink global high-speed internet project as well as a co-existing Mars rocket project.
Musk says the company needs to be “spartan” in its expenditures. The revamped approach to Raptor appears to be an attempt to cut costs by developing the rocket in stages. First it will go to the Moon and potentially function as an orbiting product around Earth. Later it will be further developed and upgraded into a rocket with the potential to explore Mars.
While Musk believes that layoffs at his day job — electric carmaker Tesla — were necessary in order to keep the company’s products relatively affordable, he thinks the SpaceX research projects are costing too much and seems to regret the SpaceX layoffs. He reportedly said during an investor call last week:
“And so, SpaceX has to be incredibly spartan with expenditures until those programs reach fruition.”
When Musk says “insane,” he doesn’t necessarily mean it in a negative way. As a businessman, he’s referring to the fact that his company is engaged in services once only provided by governments with virtually unlimited resources. When SpaceX eventually develops a space highway to Mars for mankind, they will either be the only company doing it or the only one doing it well. Untold fortunes await SpaceX at that point.
SpaceX IPO: Will It Ever Happen?
Still, SpaceX has yet to reach out for public money via an IPO. In 2017, rumors rocked the investor community: SpaceX would be launching an IPO that year. A year later, the company went for more venture capital funding, selling some shares at $135 to Fidelity. Its valuation at that time was $27.5 billion.
While SpaceX could focus on more terrestrial projects like competitive satellites, which they are also into, the company’s true mission is deep space exploration. The potential for profit from space exploration is an unknown quantity, but it could easily go into the trillions. As profit goes, one potential area of investment would be minerals acquisition. The quantities of gold, platinum, and other rare materials in space by definition outsize supplies on earth. Developing a profitable method of extraction from foreign planets and space rocks is but one way that SpaceX could, in the long run, become the most profitable company in history.
Tesla Opens A Chaotic Trading Day After SpaceX Tweets
As for Tesla, trading early this morning showed a big sell-off. Buy orders kicked in and kicked it up momentarily. After an hour or so of trading, things were on a recovery path.
While SpaceX is a separate company from Tesla, its fate is very much tied to Tesla. Good news about SpaceX can reflect well on Tesla, and vice versa.
Certainly the weekend’s activities demonstrate one thing about Musk: he is fully grounded. He understands that SpaceX needs to get realistic in its endeavors, and completion of rockets for near-term use (and potential sale to governments and others) is one of many ways they can do so. As for SpaceX engineers, this was likely not the first or last time they’ll be putting in long weekend hours in pursuit of a noble dream.
Author: P.H. Madore
Image Credit: Featured Image from Shutterstock. Charts from Tradingview.
DOVU, a London-based startup aiming to become the “the global marketplace for transport data,” is to incorporate cryptocurrency earning as part of its latest project.
The company has always had strong support since its launch two years ago backed by seed funding from InMotion Ventures, Jaguar Land Rover’s investment arm, and Creative England — a fund backed by the government. DOVU is a blockchain powered reward platform that lets users earn tokens for sharing data or changing the way they travel, then spend them on mobility-related products or services.
Irfon Watkins, company founder wants to incorporate blockchain solutions and data tokenization within its marketplace for transport data, alongside working with other companies such as vehicle hire, insurance companies, and ridesharing operations in order to create a network of pooled data.
The startup will now begin work with the rail company to improve the experience for their rail customers. Go-Ahead will benefit from DOVU’s blockchain-powered reward platform by gaining more knowledge about passenger behaviour. The first arena for the new project will be Go-Ahead’s Thameslink and Southern Rail services.
Now participating travellers will be able to earn crypto as they travel by sharing such data with the company as a type of loyalty reward for using the service.
After Ethereum core developers expressed broad support in January for ProgPoW—an ASIC resistant upgrade to the Ethereumnetwork—the Ethereum community raised concerns that the highly anticipated patch may cause the same problem it was intended to solve. Now the future of ProgPoW is in question.
ProgPoW is short for Programmatic Proof of Work. The proposed upgrade would extend Ethereum’s existing Proof of Workalgorithm, slightly altering the structure of math problems that mining nodes are tasked with solving. The change is meant to substantially improve the odds for GPU miners over ASIC miners.
Ethereum’s Pledge of ASIC Resistance
ASIC is short for “application specific integrated circuit.” In the context of Ethereum, it’s a highly specialized computer designed to be mass produced for the sole purpose of miningEthereum. However, this issue existed before Ethereum. The original Ethereum white paper called attention to the proliferation of ASICs in Bitcoin mining. Seen as a cause behind the centralization of the Bitcoin network, the Ethereum white paper set an intention to avoid the same fate:
“This means that Bitcoin mining is no longer a highly decentralized and egalitarian pursuit, requiring millions of dollars of capital to effectively participate in.”
ASICs built for Ethereum have anywhere from a two to four-fold advantage over GPUminers. Multiple Ethereum ASICs hit the market last year with the Antminer E3 from mining hardware giant Bitmain at the high end of the performance scale. The ProgPoW update is engineered specifically to slightly alter the problem that miners are asked to solve in a way that will reduce or eliminate the advantage ASICs have over GPU miners.
Why Ethereum Prefers GPUMining
The Ethereum community generally agrees in ASIC resistance, believing that GPUmining leads to a more distributed network and that the more distributed the network is, the better. While the mass manufacturing of ASIC miners provides a cost-saving advantage per hash, these miners are generally purchased in large quantities for mass-scale mining operations that consolidate ownership of the network. In addition to centralizing the network, the homogeneity resulting from running all of the same mininghardware at scale leaves the network at risk when a security vulnerability is discovered in a popular line of ASICs.
Alternatively, GPU miners are generally custom-built entirely from personal computing hardware already available in the consumer market. These machines use computer chip architecture common in deep learning and animation rendering that combines a series of 3D video cards with high-end GPU chipsets in a single machine.
The high level of accessibility of the hardware required to build these machines creates an opportunity for more people to participate and a more distributed network. As the markets change for 3D video cards, new mining machines use different combinations of hardware from different manufacturers in an attempt to fine-tune a better machine. The multi-purpose nature of the hardware also means that if markets change and miningbecomes less profitable, GPUs can be removed from the network and used in other ways. Meanwhile, ASICs serve no other purpose when they are no longer profitable to run.
The ProgPoW Debate
At the Jan. 4th core developers meeting, progress on ProgPoW was reportedly on track and running on the Gangnam test network. Since then, the Ethereum community is reporting that the ProgPoW update may lead to more centralization of the network, rather than less.
GPU miners tend to utilize hardware from one of two primary chip manufacturers, AMD and Nvidia, both of whom have reviewed the ProgPoW code. However, according to a variety of reports, the ProgPoW patch seems to provide a significant advantage to Nvidia based GPUs over AMD counterparts. To add to the controversy, one of the three team members working on the patch worked previously as a hardware engineer at Nvidia, leading to rumors of Nvidia’s involvement in the patch.
Others have been quick to follow up with claims that the patch isn’t really needed to begin with and have questioned the wisdom of altering the miningalgorithm if the intention is to move to a Proof of Stake alternative soon. Just last year, Ethereum’s creator Vitalik Buterin downplayed the ASIC threat.
Fearing that the community reaction would lead to ProgPoW not being released at all, one member of the Ethereum community is threatening to hard forkEthereum to enable ProgPoW.
The team working on ProgPoW, known online as IfDefElse posted a public response to the controversy in the form of an FAQ. In the FAQ, the team openly describes their collaboration with both Nvidia and AMD. The post goes on to clarify—in extreme detail—a technical explanation of the change and why the community experienced the variance in results.
“It was designed to have as level of a playing field as possible.… Performance of a GPU for ProgPoW in mining workloads will reflect the average gaming performance of that GPU.”
The team also addressed the possibility that ASIC manufactures will create another series of machines and the cycle will start over. The team believes that, at best, a new generation of ASICs would not be able to achieve more than a 1.5x advantage over GPUmining systems. Combined with the looming switch to Proof of Stake, this may dissuade manufacturers from attempting to challenge the team’s estimate.
ProgPoW Continues to Progress Without Clarity
Even though testing is already underway, it is still too soon to know when, how, or even if ProgPoW will be released. Despite the heated online debate, the ProgPoW patch continues to progress as planned. That said, as demonstrated by Ethereum’s recent hard fork delay, development schedules are always subject to change.
At the Ethereum Core Developers February meeting, the lively discussion explored concerns ranging from test coverage and specification readability to how to make a decision on releasing the patch. Some of the developers felt the team was ready to make an implementation decision. As embodied in a comment by core developer Greg Colvin, “we can just make it, it’s our job!”.
Meanwhile, other developers suggested that the decision to upgrade may ultimately rest with the community itself. The team put a lot of emphasis on doing the right thing for the mining community. Another core developer, Piper Merriam said:
“I’ve largely stayed out of the discussion…. It’s a decision that, frankly, I’m not excited about making for the network… but I also acknowledge that it’s potentially our responsibility to make it.”
Without complete details on how the patch will proceed, the team has agreed to organize a third-party audit to verify the fairness of the new algorithm while development on ProgPoW continues as planned. Whether Ethereum will make the switch to ProgPoW or focus on implementing PoS is still highly uncertain.
A COLLECTIVE OF BOFFINS from the Massachusetts Institute of Technology (MIT) have developed a robot that can teach itself the complex physics of Jenga.
According to the researchers, the manipulator arm of the robot relies on a machine learning algorithm, visual data, and tactile feedback to teach itself how to correctly move blocks in the classic block-stacking game.
Playing Jenga requires a soft touch to prevent the tower from tumbling down. It also demands mastery of other skills, such as pushing, pulling, probing, placing, and aligning the blocks in the tower.
In its latest research, the team equipped an industrial ABB IRB 20 robotic arm with a soft-pronged gripper, an external camera, and a force-sensing wrist cuff. To train the machine, it was directed to randomly select a block in the Jenga tower and also select a location on that block to push and move it.
Each time the arm pushed the block, a connected computer recorded the force and visual measurements and compared those to previous moves before categorising the attempt as successful or unsuccessful.
Rather than performing thousands of such attempts, the robot was trained on just 300 moves. The team grouped attempts of similar measurements and outcomes in different clusters, each signifying specific block behaviour. Within around 300 attempts, the robot learned to intelligently anticipate its moves, guessing which blocks would be more difficult to move than others, and which might cause the tower to collapse.
The performance of the robotic arm was also compared with the performance of human players. The team found the success rate of the robot in keeping the tower upright while removing the wooden blocks was almost on a par with that of human players.
The team believes this technology could be useful in production units where delicate touch and careful vision are needed, for example, assembly of cellular phones or other smaller parts in a, separating recycling items from waste, etc.
“In a cellphone assembly line, in almost every single step, the feeling of a snap-fit, or a threaded screw, is coming from force and touch rather than vision,” says Alberto Rodriguez, Assistant Professor in the Department of Mechanical Engineering at MIT, and the lead researcher of the study.
“Learning models for those actions is prime real-estate for this kind of technology.”