Starship Robot Delivers Groceries [VIDEO]

A while back Microsoft bought out Skype for an astounding $8.5 Billion. Ever wonder what the founders of Skype are doing with all that money? They figuring out to use Unmanned Ground Vehicles (UGV) to deliver groceries. Seriously.

Skype co-founders Ahti Heinla and Janus Friis have built a small delivery UGV, and slapped it with the glorious label “Starship.” It goes about 4 MPH, carries two grocery bags (20lbs), and needs to be recharged every 30 minutes. It has a full compliment of cameras, GPS, gyroscopes, and on-board mapping data. While it has autonomous capabilities, it can also be remotely operated by a human.

We here at AMREL are always interested in new applications for UGVs, since we make Operator Control Units (OCU) for them. As much as we wish this enterprise to succeed, it’s hard to see how. For one thing grocery delivery services have a nasty habit of going belly up. Forbes even ran an article titled 10 Reasons Why Online Grocery Shopping Is Failing.

The Starship UGV travels by sidewalks. Is that even legal in most cities? What about the neighborhoods that do not have sidewalks (a fairly common occurrence in rural areas and the Western US)?

[layerslider id=”31″]

We here at AMREL are always interested in new applications for UGVs, since make Operator Control Units for them. Mostly, the economics make no sense of all. The Starship UGV, with its advanced sensor package and autonomous capabilities, must be a relatively expensive machine. With supermarkets being notorious for low-profit margins, what kind of return on investment can be expected? Factor in the fact that the Starship can only make one delivery at a time, and that it’s slow speed and short battery life gives it a very limited operating range,  this little UGV is looking less practical all the time.

Also, I think a lot of hormonally challenged young men might decide that the UGV looks more like a soccer ball than a Starship, and act accordingly.

To be fair, this service is being tried out in England, which may make more business sense. People there shop more frequently than Americans and tend to live closer to small neighborhood stores. Also, they may be less prone to vandalize helpless, innocent UGVs (although I wouldn’t count on that).

I think that there is a future for UGVs in home delivery. I could easily imagine a delivery truck acting as a mothership for a a number of UGVs delivering dry goods in a particular neighborhood.

But groceries? With an aging population, there is a real need for such a service, but I think supermarkets would be better off hiring a teenager with a bicycle or a car.

To learn more, watch video below.

Six Best Robots at IROS 2015 [VIDEO]

Recently, the 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) took place in Hamburg, Germany. The Institute of Electrical and Electronics Engineers (IEEE) posted a video of the most interesting unmanned systems at the show (see below).

In the video, most robots had recognizable human-like characteristics. This is not inevitable; some unmanned systems, such as Unmanned Ground Vehicles (for which AMREL builds Operator Control Units), do not remotely resemble humans. In fact there has been a controversy over how human-like unmanned systems should be (see Walk n’ Roll).

As evidence by the video, developer sentiment seems to be leaning towards human-like (or “humanoid”) robots. Usually, the proponents of humanoid robots offer 3 rationales:

  • Humanoid robots are better suited for performing human-like tasks
  • Humanoid robots are better suited for working in an environment built for humans.
  • Humanoid robots are better suited for Human Robot Interactions (HRI)

However, there is yet another reason why people are building robots. To explain why, let me tell you about a Psychology of Language class that I attended many eons ago. The central question of the class was “Could we teach a computer to read?” No one knew if this was even theoretically possible. At the time Optical Character Recognition was primitive at best. We weren’t even certain if a computer could transform vocal input into written text.

None of the people in the class were software engineers. We took up the challenge, not so much as to get a computer to read, but to better understand how people read. By teaching machines to read, we were forced to analyze and closely examine human strategies for reading. Most class time was devoted to breaking down the act of reading into mechanized processes, subtasks, operations, and flow charts.

[layerslider id=”31″]

A more familiar example might be the oft-repeated observation that an unmanned system is incapable of distinguishing between a tomato and an apple. In order to teach a robot how to accomplish this discernment, we have to first figure out how people do it.

Some of the unmanned systems in the video seem destined for “social welfare” activities, i.e. helping the elderly, assisting the disabled, etc. Others are obviously designed not for specific tasks, but as means to better understand human behavior. What makes a baby “baby-like?”  If you build a baby-like robot, one is forced to figure this out.

Turns out, unmanned systems not only assist us with the “dirty, dangerous, and disgusting” tasks, but also help us understand ourselves.

Triumph of Human like robots

Robot smartphone [VIDEO]

This gives a whole new meaning to saying you have an Android  phone. Japanese have taken their love for human-like robots to a whole new level, and have built “RoBoHoN,” a smartphone that is also a robot.

It speaks, dances, displays your email, projects images, and will help you remember to buy toothpaste. It uses voice and face recognition to aid in its interactions with humans. And yes, it has an Android OS.

RoBoHoN is the latest in a string of victories for proponents of human-like robots. While having a face similar to a person my facilitate Human Robot Interaction (HRI), there is no practical reason for RoBoHoN’s legs and arms.

[layerslider id=”31″]

Actually, from an ergonomic point of view, a human-shaped smartphone is a catastrophe for the traditional way a handheld communication device is used. Of course, the makers of RoBoHoN may be counting on people adjusting their smartphone behavior to accomidate the new form factor. Why wold anyone want to go to all that trouble? It’s just so darn cute! Cuteness über alles!

As expressed on the RoBoHoN website:

A phone in human shape

A phone that you feel like talking to

A phone that also wants to know you

To hear what you hear

To see what you see

To share the same dreams

That is little RoBoHoN’s big dream

Don’t you want to take RoBoHoN with you everywhere?

I’m not so sure. Do I really want a little doll to wake me up in the morning and follow me around all day?  It seems to me there is a fine line between cuteness and creepy.

Make up your own mind by watching the video below:

Three Surprising Applications for Robots [VIDEOS]

Everyone is looking for the “killer app,” the must-have application that will speed the consumer adoption of unmanned systems.  As evidence by the following videos, the next big thing may be something that none of us thought of.

[layerslider id=”31″]

Israeli Unmanned Developers are FAST

The article below, originally published in Strategy Page, details the development, deployment, and sale of the Israeli-made Micro Tactical Ground Robots (MTGR). It illustrates several key reasons why tiny Israel is a giant in the unmanned world:

  • Close cooperation between military and civilian institutions. During the 2014 war with Hamas, the military authorities identified a need for tunnel-traveling Unmanned Ground Vehicles, and immediately informed local manufacturers.
  • Speed. Within a month, unmanned suppliers produced a design, got it accepted, and then developed, manufactured, and delivered working models. This is quite a contrast to the slow pace of the struggling American AEODRS program.
  • Leveraging obtainable technologies. Before the deployment of the MTGR, no one had an unmanned system capable of dealing with the complex and dangerous Hamas tunnels. Rather than focus on new (and presumably profitable proprietary solutions), the MTGR simply adapted existing technologies.
  • Small is good. This blog has previously noted the popularity of smaller ground robots. However, in this case “small” also refers to the country. Famous for its astronomical size, America’s defense sector has been criticized for its redundancy and endless bureaucracy. In contrast, virtually everyone in Israel knows each other. When the discovery of the Hamas tunnels sent a jolt of fear through Israeli society, the MTGR developers undoubtedly felt it as well. Many probably had family fighting in Gaza. In Israel, war is not something that happens to someone else. Desire for profit played a role in the manufacturers rush to create the MTGR, but they were also deeply concerned about protecting their communities and soldiers from terrorists.
  • Combat tested. The Technical Readiness Level (TRL) of 9 is highly valued by the American military. It means that the item under consideration has been proven in the real world.  Israel, unfortunately, has lots of opportunity to test defense systems in actual combat.

I am not sure how we can adapt the Israeli model to the US Defense sector.  Whatever frustration we may have with the endless attempts at Defense procurement reform, we can at least be assured that developing effective Defense systems in a timely matter is possible.


Infantry: Robots Hurry Up And Evolve

An Israeli firm has managed to sell some of its small battlefield robots to the American military, which is a first for an Israeli firm. The U.S. Air Force has ordered over 200 Israeli MTGR (Micro Tactical Ground Robot) for their bomb disposal teams. This came after MTGR demonstrated its capabilities during the 2014 war with Hamas in Gaza. This was particularly true with the large number of Hamas tunnels discovered. These proved more complex and dangerous than any previously encountered and a new robot was needed to deal with the situation. Within days a specification was provided to Israeli robot manufacturers and by the end of July 2014 a new robot design had been accepted, in production, delivered and in action. This was the MTGR and while it was not a major breakthrough, it was simply a better application of design elements that had been developed since the 1990s and suited current Israeli needs. The Israelis have ordered over a hundred MTGRs for delivery ASAP. Based on its success in Gaza MTGR is being offered to other armed forces and police departments around the world.

MTGR is a 7.3 kg (16 pound) tracked (or wheeled version weighing 9 kg) robot. Tracks are preferred for climbing stairs and getting over obstacles. MTGR can carry up to 9 kg of accessories. The basic MTGR comes with five cameras, a microphone, and can carry additional sensors. The cameras have day/night capability, 360 degree views and x10 zoom. One of the more useful accessories is a robotic arm for clearing debris or searching. Another useful item are bright LED lights when you need illumination. MTGR uses GPS and can carry a laser rangefinder to measure dimensions of where it is. The battery lasts 2-4 hours depending how onboard equipment is used. Top speed is 50 meters a minute and max range for the operator is 500 meters.

[layerslider id=”31″]

MTGR is designed to be carried in a backpack and an operator can use the handheld control unit to operate several MTGRs at once. The MTGR was a lifesaver for exploring Hamas tunnels, which are often filled with booby traps and other nasty surprises for advancing Israeli troops. Often an MTGR was simply sent down, take a lot of measurements and pictures and then withdraw after which explosives will be lowered down and the tunnel collapsed. If MTGR detects documents or electronic devices like laptops, tablets or cell phones, MTGR will carefully survey the area and troops will go down to recover the valuable intel often found on such devices. If MTGR can reach cell phones or small tablets it can pick them up and carry them away.

What made MTGR special was the firm that provided it demonstrated that it was able to take existing technologies and quickly adapt them to new situations. The small firm that developed MTGR it had an existing design modified and readied for production in less than a month. In wartime this is a very valuable capability. This has now been demonstrated under combat conditions and the rest of the military robot industry has to adapt.

The U.S. Army has been using robots like the MTGR since the 1990s. American designs went through rapid refinement after September 11, 2001 because thousands of these robots were bought and used by American troops in combat. The culmination of all that was expressed in the XM1216 SUGV (Small Unmanned Ground Vehicle). SUGV was designed to be the definitive next generation infantry droid, replacing existing droids like the similar but larger PackBot. Not surprisingly MTGR is based on the same experience but more refined and using some newer technology.  This design was not ready for action until most of the fighting in Iraq and Afghanistan was over. Thus by 2012 only about 200 of these combat robots were in service or on order. It was only in 2011, after more than six years of development, that the army bought its first production model SUGV. Many in the U.S. Army were not satisfied with how long it took to get SUGV to the troops and MTGR is proof that it could have been done faster.

Before September 11, 2001, the army didn’t expect to have robots like PackBot or SUGV until 2013. But the technology was already there, and the war created a major demand. The robots expected in 2013 were to be part of a new generation of gear called FCS (Future Combat Systems). SUGV is still waiting for some of the high tech FCS communications and sensor equipment (which MTGR used), and appeared in 2011 using off-the-shelf stuff in the meantime. The troops don’t care, as long as it worked. These small robots have been quite rugged, having a 90 percent availability rate.

The overly ambitious, expensive and much delayed FCS program was cancelled in 2009 but successful bits, like SUGV, were allowed to keep moving. This was a big deal for SUGV, because demand for these small droids collapsed when the Islamic terror offensive in Iraq did in 2008. There were plenty of droids left over for service in Afghanistan, where the Taliban provided a much lower workload for the little bots than did Iraq.

SUGV is a 13 kg (29 pound) robot, similar to the slightly older and larger Packbot. SUGV can carry 3 kg (6.6 pounds) of gear, and seven different “mission packages” are available. These include various types of sensors and double jointed arms (for grabbing things.) SUGV is waterproof and shock resistant. It fits into the standard army backpack, and is meant to operate in a harsh environment. The battery powered SUGV is operated wirelessly, or via a fiber optic cable, using a controller that looks like a video game controller with a video screen built in. SUGV can also use an XBox 360 controller, with the right drivers. Like the earlier PackBot and later MTGR, SUGV can climb stairs, maneuver over rubble and other nasty terrain.

The SUGV design is based largely on feedback from combat troops. For example, it is rugged enough to be quickly thrown into a room, tunnel or cave, activated and begin sending video, as well as audio, of what is in there. This feature makes it very popular with the troops, who want droids with the ability to see, hear and smell were more acutely. No one likes being the first one going into dark, potentially dangerous, places. Throwing a grenade in first doesn’t always work, because sometimes frightened civilians are in there. Despite all these fine qualities, the current generation of robots is not fast enough, agile enough or sensitive enough to compete with human troops doing this kind of work. Sometimes, however, the robots are an adequate, and life-saving, substitute. SUGV is supposed to be better at this sort of thing.

SUGV can also perform outpost and listening post work. These are two dangerous jobs the infantry are glad to hand off to a robot. Outposts are, as the name implies, one or two troops dug in a hundred meters or so in front of the main position, to give early warning of an enemy attack. A listening post is similar, but the friendly troops are often much deeper into enemy territory. The SUGV battery enables it to just sit in one place, listening and watching, for eight hours or more. After that, you send out another SUGV with a fresh battery, and have the other one come back for a recharge. No risk of troops getting shot at while doing the same things, and the troops really appreciate that. Again, the problem with this is that the robot sensors are just not there yet. The sensors are getting close, but not close enough for troops to trust their lives to this thing.

Other dangerous jobs for the SUGV are placing explosives by a door (to blow it open for the troops), or placing a smoke grenade where it will prevent the enemy from seeing the troops move. Since 2006 users of the older PackBot UGVs filled military message boards with interesting uses they have found for these robots, and new features they could make use of. SUGV and MTGR are the products of all that chatter.

 [layerslider id=”31″]

There Are No Robot Surgeons

Recently, I sat down to talk with Daniel Naftalovich, who is a PhD candidate in robotics engineering & control systems at California Institute of Technology. Simultaneously, he is pursuing a medical degree from the University of Southern California (he claims he has time to sleep, but somehow I doubt it). As may be expected, he had a great deal to say about the use of unmanned systems for use in surgeries.

“The one thing I try to make people understand is that robot surgeons do not exist,” says Naftalovich.

This assertion may seem strange, because we hear a lot about the increasing role of unmanned systems in the healthcare field. Indeed, a Google search yields over a million results for “robot surgeon.”

What actually is happening is robotic-assisted surgery. In my discussion with Naftalovich I was surprised to learn that surgical robots have virtually no autonomy. “Robotic surgeons” are really computerized tools, which enable the human doctor to teleoperate. Unlike traditional surgery, in which the doctor stands on his feet, hovering over the patient, robotic assistance enables the surgeon to sit down a few feet away, often in the same room.

[layerslider id=”31″]

“Teleoperated surgery offers a number of advantages,” explains Naftalovich. “For one thing, a surgeon who is sitting has greater endurance during lengthy operations than one who is standing.” The physical toll on surgeons during long operations is a real problem; neck problems are a prominent career ender for many surgeons.

Teleoperation offers another advantage. More surgical tools can be placed above the patient, since the doctor is no longer above him. “A robot can have four arms, while a human has only two,” says Naftalovich. This allows a more rapid transition between tools during operations, and better tool management.

Often in Human-Robot Interactions (HRI), the human has to learn unnatural motions in order to use an unmanned system to execute an action. For example, to get an effector arm to grasp and pick up an object, a human operator may have to maneuver a joystick or even punch codes into a keyboard.

In contrast, robotic-assisted surgery allows human doctors to be more human, i.e. use more natural motions to execute tasks. Most robotic-assisted surgery has been focused on laparoscopy, which reduces the size of an incision by inserting a thin, lighted tube into the patient’s body. Naftalovich describes a typical traditional laparoscopic tool as resembling “a pair scissors tied on the end of a stick.”  Since the tool rests on a fulcrum, the doctor has to adjust his movements accordingly. When he moves his hands up, the effector arm moves down. When he moves his hand to the right, the tool goes to the left, and so on.  Robotic-assisted surgery eliminates this inverted motion and allows the doctor to move in a normal, more intuitive way.

Robotic-assisted surgery can enhance motion as well. It can eliminate the tremor of a doctor’s hand (a real problem in lengthy surgeries). In addition, it can scale the doctor’s movement to a size that is appropriate. It evens allows the surgeon the surgeon to rotate the effector arm with his wrist motion, something traditional tools will not do. What robotic-assisted surgical tools cannot do is operate autonomously.

[layerslider id=”31″]

I suggested that autonomous surgical capabilities may be added piecemeal, as they are with automobiles. Some new cars have autonomous capabilities, such as anti-collision features and parking. Similarly, a relatively simple autonomous capability, such as suturing, could be added to robotic-assisted surgical tools. Even though simple procedures, such as robotic suturing, are being attempted in laboratory settings, Naftalovich believes there is a long way to go until they are integrated into a surgical workflow.

“I think about autonomy all the time,” says Naftalovich. “But we are just not there yet.” Currently, robotic-assisted surgical tools only enhance the surgeon’s capabilities; they do not replace him.

Currently, robotic-assisted surgeries are primarily laparoscopic abdominal operations. Naftalovich is working on control systems that would allow robotic tools to be used in neurosurgeries. The primary challenge is integrating microscopic capabilities.

Civilian IED Deaths

While researching an article on the Defense budget, I came across these two charts.

troop levels


Source: Council of Foreign Relations

Intellectually, I knew that we had dramatically scaled down our involvement with Iraq and Afghanistan, but I didn’t fully appreciate how much until I saw these two charts. Sure, we’re still very much involved in the Middle East, and we may still get drawn back even further, but as of right now, major American combat operations in that part of the world are a shadow of what they were.

While it is great news that our soldiers are not in harm’s way as much as they were, those involved in maintaining counter-IED capabilities may be wondering about their fate. Are counter-IED personnel, programs, and equipment to be put on the shelf until the next land war? Joint Improvised Explosive Device Defeat Organization (JIEDDO) has expressed concerned about its ability to preserve the valuable skills developed during recent conflicts.

This is a problem that affects other capabilities besides counter-IED. During peace time we can’t maintain a factory producing a wartime-level of helicopters. However, if war breaks out, we don’t want to waste time training workers, developing supply lines, and building the factory. NDIA has a nice summary of the challenges of Maintaining a Viable Defense Industrial Base.

[layerslider id=”31″]

Those involved in counter-IED efforts may not face this dilemma. IEDs are a civilian, not just a military problem. Public safety and security agencies will need assistance from those who have counter-IED skills and equipment. The article below, which is excerpted from Fierce Homeland Security, discusses this phenomenon.

Improvised explosive devices aren’t just a significant threat to soldiers on the battlefield, but they’re also a leading killer of civilians worldwide, Interpol’s secretary general said this week in a news interview in Australia where the first-ever international forum to counter such threats was held.

“For instance, we have had more than 10,000 civilian casualties just this year around the world,” said Jürgen Stock, during a Sept. 3 radio interview with the Australian Broadcasting Corp.

He said IEDs are the “weapon of choice” for terrorists and other criminals and a leading killer of civilians. “So it’s really a global threat and what we need is a better information sharing between the military and law enforcement,” he added.

Almost three weeks ago, an improvised explosive device, or IED, killed 20 people at a popular Hindu shrine in central Bangkok, Thailand. More than two years ago, two brothers detonated an IED during the Boston Marathon that killed three people and injured hundreds more.

Stock said that the Internet has made it easier for terrorists and others such as so-called lone wolves – individuals acting alone in attempting to or committing terrorist acts – to put together such bombs.

Interpol, the international criminal police organization that’s based in France, co-hosted the Sept. 2-4 international conference along with Australian law enforcement and defense agencies to develop or enhance information sharing about IEDs and training capabilities among police and military personnel.

In the radio interview, Stock also talked about the role his organization is playing against foreign terrorist fighters. Many governments worldwide are concerned about their nationals traveling to conflict zones in Syria and Iraq to join the terrorist fight there as well as learn bomb-making skills that they can use to commit terrorist acts when they return to their home countries.

Stock said there are currently 25,000 such fighters from more than 100 countries that have joined the conflict zones and terrorist groups like the Islamic State of Iraq and Syria.

He said Interpol is building global platforms and databases to share information about these individuals with its 190 member states. One database contains 5,000 profiles of these fighters while there are other databases that include fingerprints, DNA material, and lost, stolen and forged identity or travel documents such passports.

He said it’s important that these databases and tools are available and accessible to border security agents and police agencies.

 [layerslider id=”31″]

First Licensed Autonomous Truck [VIDEO]

The first officially licensed autonomous-capable semi-truck is here. Licensed in Nevada (only on highways during daylight hours and good weather), it is not a driverless vehicle; humans are still required. However, the human is not an operator, at least all the time. In theory, he could be playing Candy Crush or watching a ball game.

The truck is named “Inspiration.” Freightliner, its manufacturer, is by far not the only one looking to make a splash in the autonomous car game. Apple, Tesla, Uber, and of course ,Google all have plans for autonomous cars.

Toyota, working with MIT and Stanford, will invest at least $50 million over the next five years in “assisted autonomy.”  Unlike the famous Google car, this will not be a fully autonomous vehicle. Toyota’s goal is to have AI that prevents crashes of human-operated cars. This neatly avoids many (but not all) legal liability issues that plague autonomous systems.

[layerslider id=”31″]

In spite of all the interest of big players, Freightliner, is the first licensed autonomous vehicle. This is the first vehicle that is legally allowed to operate autonomously. This means on your next visit to Las Vegas, you might see a big semi barreling down the highway, while the person at the steering wheel is reading a newspaper.

Learn more about the Inspiration at IEEE Spectrum or check out Freightliners cheery video below.

This video explains how radar, cameras, and “platooning” work.

Space Robot’s Unique Locomotion [VIDEO]

Scientists at the Jet Propulsion Laboratory (JPL) faced an unusual problem. NASA wanted a rover that could navigate the uneven surfaces of asteroids and comets. However, rovers, such as the ones used on Mars, have treads and wheels. In the low gravity environment of small extra-planetary objects, the rovers with wheels would lose their grip on the ground and literally drift off into space. None of the conventional means of unmanned propulsion – propellers, wings, or wheels –would work. How would you move an unmanned system when it can’t swim, walk, fly, or roll?  Their solution, as demonstrated in the video below is ingenious.

You can also learn more at

[layerslider id=”31″]

Happy Anniversary Curiosity Mars Rover! [VIDEOS]

It’s the anniversary of the Curiosity’s landing.  For three years, the Mars rover has successfully navigated the rough terrain of the red planet, looking for water and signs of life.

We should take a moment to remember just how improbable was the success of the landing. For one thing, the fourth planet had proved not especially hospitable to early missions. In fact, Mars had been nicknamed the “Probe killer.”

Here are a few facts and figures about the Curiosity’s landing on Mars.

154 million miles

Distance of Earth from Mars at time of landing.

7 minutes                

Time delay for communication from Mars to Earth.  This is a remote controlled vehicle, so commands take 7 minutes to reach Curiosity, a significant limitation. “7 minutes of terror” is the nickname for the delay from the time of the landing to the moment that the rover’s signal reached Earth

1600 degrees

Heat of decent.

100 times thinner than Earth’s

Thickness of Mars skimpy atmosphere.

Thousand miles an hour

Speed of decent before parachute is deployed.

6,500 pounds of force

Strength necessary for the parachute.

100 pounds

Weight of parachute.


Force generated by deployment of parachute.


Vehicle configurations.


Pyrotechnic devices.


Lines of code used.


Margin of error

Keep in mind that there was an important onboard package of instruments that had to be protected at all costs, including:

  • 17 cameras
  • Alpha Particle X-ray Spectrometer
  • Chemistry and Camera
  • Chemistry and Mineralogy
  • Dynamic Albedo of Neutrons
  • Mars Descent Imager
  • Mars Hand Lens Imager
  • Mast Camera
  • Radiation Assessment Detector
  • Rover Environmental Monitoring Station
  • Sample Analysis

The above considerations and others necessitated a Rube Goldberg landing system that looks like a particularly implausible stunt out of a James Bond movie.  To see the bizarre strategy adopted for landing, watch video below.

[layerslider id=”31″]

For an update of Curiosity’s current status, watch video below: