Going Soft On Robots

Snake on a plane! Don’t panic, it’s probably just a (soft) robot

My piece on soft material robots for Reuters. Original story here: I’ve added links in this piece.

SINGAPORE (Reuters) – Robots are getting softer.

Borrowing from nature, some machines now have arms that curl and grip like an octopus, others wriggle their way inside an airplane engine or forage underwater to create their own energy.

This is technology that challenges how we think of, and interact with, the robots of the not-too-distant future.

Robots are big business: by 2020, the industry will have more than doubled to $188 billion, predicts IDC, a consultancy. But there’s still a lot that today’s models can’t do, partly because they are mostly made of rigid metal or plastic.

Softer, lighter and less reliant on external power, future robots could interact more safely and predictably with humans, go where humans can’t, and do some of the robotic jobs that other robots still can’t manage.

A recent academic conference in Singapore showcased the latest advances in soft robotics, highlighting how far they are moving away from what we see as traditional robots.

“The theme here,” says Nikolaus Correll of Colorado University, “is a departure from gears, joints and links.”

One robot on display was made of origami paper; another resembled a rolling colostomy bag. They are more likely to move via muscles that expand and contract through heat or hydraulics than by electricity. Some combine sensing and movement into the same component – just as our fingertips react to touch without needing our brain to make a decision.

These ideas are already escaping from the lab.

SMALL, AGILE

Rolls-Royce, for example, is testing a snake-like robot that can worm its way inside an aircraft engine mounted on the wing, saving the days it can take to remove the engine, inspect it and put it back.

Of all the technologies Rolls-Royce is exploring to solve this bottleneck, “this is the killer one,” says Oliver Walker-Jones, head of communications.

The snake, says its creator, Arnau Garriga Casanovas, is made largely of pressurized silicone chambers, allowing the controller to propel and bend it through the engine with bursts of air. Using soft materials, he says, means it can be small and agile.

For now, much of the commercial action for softer robots is in logistics, replacing production-line jobs that can’t yet be handled by hard robots.

Food preparation companies and growers like Blue Apron, Plated and HelloFresh already use soft robotics for handling produce, says Mike Rocky, of recruiter PrincetonOne.

“This is an area robots traditionally can’t do, but where (soft robots) are on the cusp of being able to,” said Nathan Wrench of Cambridge Consultants.

MARINE INSPIRATION

Investors are excited, says Leif Jentoft, co-founder of RightHand Robotics, because it addresses a major pain point in the logistics industry. “Ecommerce is growing rapidly and warehouses are struggling to find enough labor, especially in remote areas where warehouses tend to be located.”

Some hope to ditch the idea that robots need hands. German automation company Festo and China’s Beihang University have built a prototype OctopusGripper, which has a pneumatic tentacle made of silicone that gently wraps itself around an object, while air is pumped in or out of suction cups to grasp it.

OctopusGripper (photo: festo.com)

A soft robot fish from China’s Zhejiang University swims by ditching the usual rigid motors and propellers for an artificial muscle which flexes. It’s lifelike enough, says creator Tiefeng Li, to fool other fish into embracing it as one of their own, and is being tested to explore or monitor water salinity.

And Bristol University in the UK is working on underwater robots that generate electrical energy by foraging for biomatter to feed a chain of microbial fuel-cell stomachs. Hemma Philamore says her team is talking to companies and environmental organizations about using its soft robots to decontaminate polluted waterways and monitor industrial infrastructure.

This doesn’t mean the end of hard-shelled robots.

Part of the problem, says Mark Freudenberg, executive technology director at frog, a design company, is that soft materials break easily, noting that most animatronic dolls like Teddy Ruxpin and Furby have rigid motors and plastic casings beneath their fur exteriors.

To be sure, the nascent soft robot industry lacks an ecosystem of software, hardware components and standards – and some companies have already failed. Empire Robotics, one of the first soft robot gripper companies, closed last year.

RightHand’s Jentoft says the problem is that customers don’t just want a robot, but the whole package, including computer vision and machine learning. “It’s hard to be a standalone gripper company,” he says.

And even if soft robots find a niche, chances are they still won’t replace all the jobs done by human or hard-shelled robots.

Wrench, whose Cambridge Consultants has built its own fruit picking robot, says he expects to see soft robots working with humans to harvest fruit like apples and pears which are harder to damage.

Once the robot has passed through, human pickers would follow to grab fruit hidden behind leaves and in hard-to-reach spots.

“It’s a constant race to the bottom, so there’s a pressing business need,” Wrench said.

Reporting by Jeremy Wagstaff; Editing by Ian Geoghegan

Narrowband Goes Broad

Seems LoRa is really taking off. Citing data from research firm Analysys Mason, Chris Donkin writes that 85 new networks were announced as live, in a trial phase or in development in 2016 compared with 29 in 2015.

While early LPWA deployments were concentrated in the US and Western Europe, Analysys Mason found interest in the technology spread during 2016, with strong traction being seen in the APAC market.

During 2015, two thirds of initiatives took place in the US and Western Europe whereas in 2016 the figure was down to less than a third. Simultaneously APAC showed growth from 4 per cent in 2015 to 30 per cent in 2016.

The report identified developments in Japan, Singapore, South Korea, Australia and New Zealand as being especially significant in the regional shift identified last year.
– via Mobile World Live

While a lot of these led by SigFox or operators using the NB-IoT standard — a stripped down 3G, more interesting, I think is the LoRa version, which actually provided the single largest group — 29 deployments vs 27 Sigfox.

The LoRa Alliance says 17 nationwide deployments have been publicly announced, and there are live networks in more than 150 cities. So I’m guessing AM’s numbers are somewhat conservative. The Things Network, an open source implementation of LoRa, boasts dozens of communities — people who are working on networks, however small — and while most are in Europe and the US, Australia is strong — Sydney’s Meshed Network Pty has installed five gateways around the city.

The author of the AM piece, Aris Xylouris, says “we can expect more announcements to be made before Mobile World Congress (MWC) 2017. It is likely that the launch of the first real commercial deployment of an NB-IoT network will be among the announcements at MWC 2017.”

Here’s my take from August on narrowband.

2017 Predictions

This piece was written for the BBC World Service’s Business Daily.

This year is going to be an interesting one, but in technology it’s going to be particularly so. Social media is going to see some reverses, as users start to wake up to the compromises they make in sharing information with companies, governments and the world. But the real progress is going to be making our machines understand us better, in ways that we want.

Artificial intelligence: you’re no doubt rolling your eyes at the phrase, given how many times this technology has been promised as being around the next corner. I’m with you. But I think the focus has been on the wrong place: voice. Apple’s Siri has not been a huge success — except for my daughter, who loves talking to an adult she can be rude to — and Amazon’s Alexa, though impressive, is going to confine itself to those places where we feel comfortable talking to machines: the home.

That makes it inherently limited. Ours is actually a largely text-based world — we still use email, we prefer to text, or Whatsapp our friends, and this is where AI is going to be most useful. I already use an AI assistant called Evie to schedule my appointments; she parses emails I send her and, with a little human help, sets up meetings and calls on my behalf. I save an hour or so a week.

Expect to see more of this: using natural language – the way we usually write — to interact with devices, not via special apps but via whatever channels we already use. It’s our devices — fridges, computers, databases — that have to learn our language and preferred medium, not the other way around. AI will be a success if it can master this, and this year will be key.

Indeed, the same principle will be applied elsewhere: removing the machine-like elements of our interactions. AI will help us talk to machines better, but machines will also help immerse us in experiences. Pokemon Go, the mobile app that led many people astray catching and battling weird critters, was a hit because it took a decade-old technology, augmented reality, and bolted it onto something that people actually found useful. Well, not useful, exactly, but compelling.

Augmented reality took technology into the real world, and gave it an enticing layer. The next step — using technology to shrink the distance between people and the real world. Optimists are calling it teleportation — moving you to places you wouldn’t normally go, or can’t go. That could be a 360 degree video from a live event, or drones filming from way above you, or even experiencing something akin to physical touch with someone whose far away from you. A Singapore startup offers a remote kissing machine, which it of course has called the Kissenger.

Industry is getting excited about this because it sees the possibility of creating a digital twin of a real world device — a turbine say — and then manage and experiment on that digital version of the real thing. A Malaysian company does something similar with corpses — scanning the deceased so that post-mortems can be conducted digitally. The original body is left untouched — which may please relatives, but also means the number of post-mortems can be limitless, and performed by someone on the other side of the world.

All of this technology is available now, but it still takes some vision and money to bring it to market. But what people want is clear enough: technology should bring people closer to each other and their machines, but stay out of the way as much as possible. We may not successfully wean ourselves off our mobile screens any time soon, but we could at least make what we see, hear, and do on those screens as useful, exciting and human as possible.

PCs with Wireless Charging?

PCWorld reports of a Dell laptop with wireless charging, but it seems a low-key affair without much conviction:

At CES last week, Dell showed a wireless charging PC called the Latitude 7285, a 2-in-1 with a detachable screen attached to a keyboard base. It’s the first wireless charging laptop based on the AirFuel Alliance’s emerging wireless PC charging standard.

But Dell doesn’t have widespread plans to put wireless charging in a host of new devices. That’s partly because the technology, with slow charging speeds, is limited to low-power devices and isn’t mature enough to replace wired charging. The wireless charging Latitude 7285 has a low-power Intel Kaby Lake chip that draws just 4.5 watts of power.
– via PCWorld

You can see the problem. The whole point of wireless charging is that it works for smaller devices that you want to charge without having to fiddle with cables. It’s also a location thing: if you’re at your desk you’ve probably got a cable. But if you’re at your bedside, and want to charge your Kindle or phone overnight, just being able to put it on the nightstand and know it’s charging is elegant and appropriate.

So part of the problem here is companies foisting a ‘solution’ on a problem that doesn’t exist. The other is the continuing failure to agree on standards that work across all devices. Until that happens, don’t expect this to be a thing. As PC World says:

It hasn’t been smooth sailing for wireless PC charging. Intel had earlier taken the lead on establishing the wireless PC charging ecosystem. But the company scaled back efforts after laying off 12,000 people last year and restructuring operations to focus more on servers, internet of things, automotive tech, and other areas.

Intel was also leading an effort by AirFuel Alliance to establish the Resonant standard for wireless PC charging. AirFuel last November reconstituted a PC Task Force to drive adoption of wireless charging in PCs, with partners including Dell, Lenovo, and STMicroelectronics.

Intel also took on the job of trying to convince airports, cafes, and other locations to install wireless charging stands for laptops. But the efforts have not yet shown any tangible results.

Dog fight: Start-ups take aim at errant drones

Here’s a piece I wrote with Reuters colleague Swati Pandey about the rise of anti-drone technologies. Buckle up.

A boom in consumer drone sales has spawned a counter-industry of start-ups aiming to stop drones flying where they shouldn’t, by disabling them or knocking them out of the sky.

Dozens of start-up firms are developing techniques – from deploying birds of prey to firing gas through a bazooka – to take on unmanned aerial vehicles (UAVs) that are being used to smuggle drugs, drop bombs, spy on enemy lines or buzz public spaces.

The arms race is fed in part by the slow pace of government regulation for drones.

In Australia, for example, different agencies regulate drones and counter-drone technologies. “There are potential privacy issues in operating remotely piloted aircraft, but the Civil Aviation Safety Authority’s role is restricted to safety. Privacy is not in our remit,” the CASA told Reuters.

“There’s a bit of a fear factor here,” says Kyle Landry, an analyst at Lux Research. “The high volume of drones, plus regulations that can’t quite keep pace, equals a need for personal counter-drone technology.”

The consumer drone market is expected to be worth $5 billion by 2021, according to market researcher Tractica, with the average drone in the United States costing more than $500 and packing a range of features from high-definition cameras to built-in GPS, predicts NPD Group, a consultancy.

Australian authorities relaxed drone regulations in September, allowing anyone to fly drones weighing up to 2kg without training, insurance, registration or certification.

Elsewhere, millions of consumers can fly high-end devices – and so can drug traffickers, criminal gangs and insurgents.

Drones have been used to smuggle mobile phones, drugs and weapons into prisons, in one case triggering a riot. One U.S. prison governor has converted a bookshelf into an impromptu display of drones his officers have confiscated.

Armed groups in Iraq, Ukraine, Syria and Turkey are increasingly using off-the-shelf drones for reconnaissance or as improvised explosive devices, says Nic Jenzen-Jones, director of Armament Research Services, a consultancy on weapons.

A booby-trapped drone launched by Islamic State militants killed two Kurdish Peshmerga fighters and wounded two French soldiers in October near Mosul.

The use of drones by such groups is likely to spread, says Jenzen-Jones. “There’s an understanding that the threat can migrate beyond existing conflict zones,” he told Reuters.

ANTI-DRONES

This is feeding demand for increasingly advanced technology to bring down or disable unwanted drones.

At one end of the scale, the Dutch national police recently bought several birds of prey from a start-up called Guard From Above to pluck unwanted drones from the sky, its CEO and founder Sjoerd Hoogendoorn said in an email.

Other approaches focus on netting drones, either via bigger drones or by guns firing a net and a parachute via compressed gas.

Some, like Germany’s DeDrone, take a less intrusive approach by using a combination of sensors – camera, acoustic, Wi-Fi signal detectors and radio frequency (RF) scanners – to passively monitor drones within designated areas.

Newer start-ups, however, are focusing on cracking the radio wireless protocols used to control a drone’s direction and payload to then take it over and block its video transmission.

Singapore’s TeleRadio Engineering uses RF signals in its SkyDroner device to track and control drones and a video feed to confirm targets visually.

DroneVision Inc of Taiwan, meanwhile, says it is the first to anticipate the frequency hopping many drones use. Founder Kason Shih says his anti-drone gun – resembling a rifle with two oversized barrels, coupled with a backpack – blocks the drone’s GPS signals and video transmission, forcing it back to where it took off via the drone’s own failsafe features.

VARIED CLIENTELE

Clients, the start-up companies say, range from intelligence agencies to hotels. DroneVision, for example, helped local police down 40 drones flying around Taipei 101, one of the world’s tallest buildings and a magnet for drone users, in a single day.

In the Middle East, upscale hotels are talking to at least two companies about blocking drones from taking shots of their celebrity guests longing poolside or in the privacy of their bathrooms.

And even while the military, Jenzen-Jones says, may have the capability to bring down drones, demand is shifting to nimbler, more agile devices to cope with attacks using smaller off-the-shelf devices. “The key is looking for systems that are scalable, lightweight and easily deployable,” he said.

HEY, REGULATORS

The problem, such companies say, is that regulations on the use of drones – and about countering them – are still in their infancy. In countries like the United States and Australia, for example, drones are considered private property, and they can only be jammed by government agencies.

“Mitigation capabilities,” says Jonathan Hunter, CEO of Department 13, “are therefore limited.”

Oleg Vornik, chief financial officer of DroneShield, however, says: “This is expected to change shortly as governments start to recognise that critical infrastructure facilities such as airports need to be able to defend themselves against drones.”

In the United States, the Federal Aviation Administration is testing various counter-drone technologies at several airports.

Interest in the space will only grow.

London will next year host the world’s first two conferences on counter-drone technologies, says Jenzen-Jones. But there will also likely be consolidation.

DroneShield’s Vornik says the company has counted 100 counter-drone start-ups, and is talking to more than a dozen of them as potential acquisition targets.

It’s too early, Vornik says, to see evidence of moves to get around anti-drone technology. But Amazon.com last month tested deliveries in the UK via drones, and published a patent describing how it might defend drones from threats, ranging from a bow and arrow to signal jammers.

Why we hate video calls

Good piece in the New Scientist about why we’ve always hated video calls:

When another New York Times reporter went to Pittsburgh in mid-1971, however, he found only 33 Picturephones in operation, with just 12 able to dial outside their own buildings. Aside from impracticalities such as cost, it seemed that, against all predictions, no one actually wanted video calling. Users were more interested in seeing graphics than face-to-face video conversation. At Bell Labs, Lucky recalls that the only person who called his Picturephone was his boss, Arno Penzias. “I found it very awkward because I had to stare at him,” he says.

More than that, I think the enduring non-appeal of video is that it doesn’t start to replace talking face to face. Face to face talking is not about seeing the other person, or looking them in the eyes — it’s about non-verbal communication — gestures, body language, touching, etc. It’s also about allowing other things to intervene — movement, distraction, interaction with objects.

Video calls are exhausting, because you are trying to replace all that with just maintaining eye contact, or at least giving the appearance of remaining engaged. It’s a new form of communication, and we’ve tried and rejected it. Whenever Cisco drag me over to their HQ for some elaborate video conference I always feel it’s a waste of time, and a major overengineering of a flawed medium.

Talking on the phone, meanwhile, suits us perfectly (although I’ve come to hate it almost as much as video calling.) As George Costanza once said, after going through a phone conversation with a blind date:

George: She had to be impressed by that conversation, had to! It was a great performance. I am unbelievable on the phone. On the date they should just have two phones on the table at the restaurant, done.

Phone calls have become useful because we are able to transfer a lot of the body language and non-verbal cues into speech (and silence). We’re still working on text chat, but we’re getting there. It works — it’s not exhausting. It’s communicating what we want to communicate, and filtering out what we don’t — and not reading, at least for the most part, anything into anything else.

Evernote Makes Employee Reading of Messages Opt-in

Evernote has been through the wringer with its decision to add machine learning to its repertoire, effectively trying to pave the way to added services based on scanning the contents of users’ notes. Users were not happy, not least because Evernote made it opt-out. The settings looked like this: 

Screenshot 2016 12 15 06 23 04

Evernote has now had a change of heart, rather coyly calling it Evernote Revisits Privacy Policy Change in Response to Feedback: No longer would it implement the planned Privacy Policy changes for January 23.

“Instead, in the coming months we will be revising our existing Privacy Policy to address our customers’ concerns, reinforce that their data remains private by default, and confirm the trust they have placed in Evernote is well founded. In addition, we will make machine learning technologies available to our users, but no employees will be reading note content as part of this process unless users opt in. We will invite Evernote customers to help us build a better product by joining the program.”

It’s probably the best solution in the circumstances, but it was poorly handled, and reflected a lack of understanding, once again, of what the product is. Evernote is simply that: a place where you can store your notes forever. That needs to be paramount. Anything else needs to support that, and not undermine it. 

Users’ reaction was becaues they prized privacy and security above other layers of features and services that may arise from running semantic engines and whatnot over Evernote. And certainly doing it via opt-out, and a privacy policy that raised suspicions.

I personally would love to see more done with my notes — complex search is still poor, finding similar notes is still poor — but I need, and I’m sure I’m not alone — to be confident Evernote isn’t going to do anything weird with my stash without my permission. Especially have employees poring over them. 

Turn off location in iOS, and Uber doesn’t work

(Update: Uber say they are looking into it.) 

Buzzfeed says Privacy Advocates Want Uber To Stop Tracking Users After Rides End but Uber responds that “by offering the option of manually entering pick-up locations, the company is giving users a choice to be tracked or not.”

It quotes Kurt Opsahl, deputy executive director and general counsel at EFF, as saying that this ‘takes away a lot of the usability.’ Part of Uber’s appeal is how easy it is to open the app and let GPS pinpoint your location for a driver. ‘As you’re trying to get picked up by the side of the road, you might not know what address you’re at,’ Opsahl said. ‘I guess you could turn it on and off again…but that’s pretty clunky as well.’”

I’d agree, and have found in my tests that it’s worse than that: turn location off, and the app no longer works. 

First off, here are the options, as described in settings in iOS:

2016 12 06 18 10 15

So it’s either Always or Never. Nothing in between. Turn to Never and things not only get clunky — meaning that you’re prompted by dire warnings every few minutes, but after a day or two you start to get blank screens, like these when you try to book an Uber. 

2016 12 06 18 09 38

2016 12 06 17 10 20

I’ve reached out to Uber for an explanation. 

Xiaomi Goes Virtually Edgeless By Using Ultrasound

NewImage

Regular readers will know I’ve been looking out for this to happen for a while: the use of sound, or rather ultrasound, as a form of interface. Here’s a Reuters piece I did on it a year ago:  From pixels to pixies: the future of touch is sound | Reuters:

Ultrasound – inaudible sound waves normally associated with cancer treatments and monitoring the unborn – may change the way we interact with our mobile devices.

But the proof will be in the pudding, I reckoned:

Perhaps the biggest obstacle to commercialising mid-air interfaces is making a pitch that appeals not just to consumers’ fantasies but to the customer’s bottom line.

Norwegian start-up Elliptic Labs, for example, says the world’s biggest smartphone and appliance manufacturers are interested in its mid-air gesture interface because it requires no special chip and removes the need for a phone’s optical sensor.

Elliptic CEO Laila Danielsen says her ultrasound technology uses existing microphones and speakers, allowing users to take a selfie, say, by waving at the screen.

Gesture interfaces, she concedes, are nothing new. Samsung Electronics had infra-red gesture sensors in its phones, but says “people didn’t use it”.

Danielsen says her technology is better because it’s cheaper and broadens the field in which users can control their devices.

That day has happened. Xiaomi’s new MIX phone, Elliptic Labs says, is the first smartphone to use their Ultrasound Proximity Software:

INNER BEAUTY replaces the phone’s hardware proximity sensor with ultrasound software and allows the speaker to be completely invisible, extending the functional area of the screen all the way to the top edge of the phone.

Until now, all smartphones required an optical infrared hardware proximity sensor to turn off the screen and disable the touch functionality when users held the device up to their ear.

Without the proximity sensor, a user’s ear or cheek could accidentally trigger actions during a call, such as hanging up the call or dialing numbers while the call is ongoing.

However, INNER BEAUTY — built on Elliptic Labs’ BEAUTY ultrasound proximity software — uses patented algorithms not only to remove the proximity sensor, but also to hide the speaker behind the phone’s glass screen.

Besides eliminating the unsightly holes on a phone’s screen, Elliptic Labs’ technology eliminates common issues with hardware proximity sensors, such as their unreliability in certain weather conditions or in response to various skin colors as well as dark hair.

This is a good first step. The point here of course, for the company, is that they can push the display right to the top, which definitely looks nice (the front-facing camera, if you’re wondering, is now at the bottom.) But the use of ultrasound has lots of interesting implications — not least for how we interact with our phones. If gestures work, rather than just say they work, it will make interacting with other devices as interesting, maybe more interesting, than voice.

The Ugliness of Short Term Hacks on the Road to Wireless

Here’s a Kickstarter project to solve the problem of no audio jack which illustrates just how thorny it is: iLDOCK – charge and listen to iPhone 7 at the same time by ildockgear — Kickstarter

ILDOCK lets you use your wired headphone while charging your iPhone 7. You can also add storage via SD, TF and USB ports with Plus. 

NewImage

The problem, as some have highlighted in the comments, is that Apple rarely grants MFi status to accessories where the lightning cable doesn’t plug directly into an Apple device. In this case, one does but the external one doesn’t. Most manufacturers get around this by making the external one a microusb. I don’t mind that, in fact it helps me, but some folk aren’t crazy about it. 

There are other issues too, of course: if you have lightning headphones and want to charge, this isn’t going to help you. 

I know I’ve written before that the future is wireless, but this project, worthy though it is, merely illustrates how ugly the interim is going to look like. 

(Via Kickstarter)