Category Robots Podcast

This summer we used our Strider optimizer coupled with rapid prototyping in LEGO to refine its 10-bar linkage.

In a tiny fraction of the time it took us to refine TrotBot’s linkage in our garage, we explored numerous variations of Strider’s linkage and their trade-offs, and were able to significantly improve its performance in these categories:

Gait Efficiency. To be energy efficient, walking gaits should be smooth and not cause the robot to rise and fall – think how much more tiring it is to do lunges than it is to simply walk. The feet should also move at constant speed – try walking while changing how quickly your feet sweep across the ground and your leg muscles will quickly complain! Strider Ver 3 was optimized for both of these features, and it’s the only mechanism that we’ve tested which can walk with a 1:1 gear ratio without the LEGO motors stalling:

Rugged Terrain. Like with TrotBot, we wanted Strider to be able to walk on rugged terrain, otherwise why not just use wheels? But, mechanical walkers are blind and dumb, and they can’t lift their feet to avoid tripping on obstacles. So we designed Strider’s foot-path to mimic how you would walk blindfolded on a path crisscrossed by roots: lift your feet high and keep them high until stepping back down to the ground, like this:

Here’s a rugged terrain test an 8-leg build:

Strength. Walkers stress frames more than wheeled vehicles, so we increased the frame’s rigidity with more triangles. Also, the leverage of long legs can put a tremendous amount of force on the leg’s joints, especially when turning them tank-style. We strengthened Strider’s leg joints by sandwiching them between beams like in the image below, which also reduces how quickly friction wears down the lips of the frictionless pins.

Here’s a test of Strider’s strength with a 10 pound load:

We made these changes while keeping Strider simple – it’s still our easiest walker to build. We also (hopefully) made our building instructions more clear. We learned how bad we are at creating building instructions from Ben’s experience teaching his walking robots class last spring. In the past when users would email us about something confusing in the instructions, we’d simply add more pictures to clarify things. This backfired in Ben’s class, since more build pictures resulted in the students getting more confused, and caused them to skim thru the pictures more quickly, and to make more mistakes. As we learned, less is more!

We also separated the leg and frame instructions to make it easier to build different versions of Strider, and we designed the frame to make it easy to swap between the battery box/IR controller, or the EV3 brick. You can find the instructions here. Here’s an example build with the EV3 brick mounted underneath – but it would probably work with the brick mounted on top as well.

For the bold, we also posted a half dozen new variations of the Strider linkage for you to try building here, along with an online optimizer for the even bolder to create their own versions of the Strider linkage. We’ve also optimized Klann’s and Strandbeest’s linkages for LEGO, and you can find their online optimizers here. (jeez, I guess we’ve been busy)

Here’s a final teaser to encourage you (or your students) to take on the challenging task of replacing wheels with mechanical legs:

The Chinese kindergarten children giggled as they worked to solve puzzles assigned by their new teaching assistant: a roundish, short educator with a screen for a face.

Scientists at the U.S. Army Research Laboratory and the Johns Hopkins University Applied Physics Laboratory have developed software to ensure that if a robot falls, it can get itself back up, meaning future military robots will be less reliant on their Soldier handlers.

The Spraybot delivered inch-accurate positioning and instantaneous heading. Its autonomous navigation had built-in obstacle selection and avoidance. It was designed to both detect and identify plants and rows and deliver spot spraying.

The Spraybot delivered inch-accurate positioning and instantaneous heading. Its autonomous navigation had built-in obstacle selection and avoidance. It was designed to both detect and identify plants and rows and deliver spot spraying.

For a lot of people, the idea of trusting your life to a computer hurtling you down the road sounds foolhardy. The whole idea behind the autonomous vehicle is to eliminate human error and deliver zero street/road fatalities – awesome, but impossible.

It can't be seen with a human eye. It doesn't look anything like C-3PO or R2-D2, or even BB-8. But, nevertheless, it is a robot (all 120nm of it) and its creators from The University of Texas at San Antonio (UTSA) are now world record holders in the Guinness World Records for creating the Smallest Medical Robot.

Experts from scores of countries are meeting to discuss ways to define and deal with "killer robots"—futuristic weapons systems that could conduct war without human intervention.

How to use recurrent neural networks and LSTM to forecast cryptocurrencies price

Autonomous submersible craft could harvest a species that is damaging critical reefs and depleting valuable fish supplies

In this episode of Robots in Depth, Per Sjöborg speaks with Søren Peter Johansen from DTI about implementing robotic solutions.

Søren talks about how he got into robotics by starting to tinker with any electronic he could get his hands on. He worked in a mechanical workshop and added automation to the machines in the shop. As robots became more and more available, he then included them in his work.

Søren also discusses examples of successful human robot collaboration and how software and hardware both are essential elements of robot development.

We also get to hear about how he went to the Danish Technological Institute because he saw an opportunity to work with lots of interesting robots.

This interview was recorded in 2016.

We are excited to announce the AI Driving Olympics (AI-DO), a new competition focused around AI for self-driving cars. The first edition is going to be at NIPS 2018; the second edition will be at ICRA 2019.

The competition comprises challenges of increasing complexity for self-driving cars, from lane following to fleet management, all on the Duckietown platform.

We are excited because AI-DO at NIPS 2018 is going to be the first robotic competition at a machine learning conference, and in a way that is completely reproducible: you send your code — we run it on our robots. Or, you can get a Duckiebot for yourself through the Kickstarter run by our newly-established non-profit Duckietown foundation.

For all details and rules about AI-DO see here.

The AI Driving Olympics is presented in collaboration with 6 academic institutions: ETH Zürich (Switzerland), Université de Montréal (Canada), National Chiao Tung University (Taiwan), Toyota Technological Institute at Chicago (USA), Tsinghua University (China) and Georgia Tech (USA), as well as two industry co-organizers: nuTonomy (a self-driving car company) and Amazon Web Services (AWS).

The newsletter The Information reports Uber’s investors are pushing Uber to sell its self-drive division to some other large player. The division has, of course, been nothing but trouble for Uber, and as I have noted several times, Uber is one of the few large players in this space that doesn’t have to build their own tech. They have the #1 brand in selling rides, and selling rides is what the robotaxi business is all about.

At the same time, Uber recruited a great team (though it has lost many of them.) First they recruited many of the best around CMU for their Pittsburgh ATC headquarters. CMU and Stanford are where most of the stars of the robocar world have come from. Then they “paid” $680M in stock for Otto, which was really an acqui-hire in many ways, at least if you believe court documents. Because that stock payout required certain milestones that weren’t met, the Otto stockholders did not get their money, but Uber had to make a large stock payout to Waymo as a result of the lawsuit over Anthony Levanowski’s actions.

So they have this team, but the team and the Uber name are tarnished by the fatality. But the right pickings from the team are still valuable to somebody who is falling behind, if you eliminate the factors that led to the fatality and take the Uber name off it.

At the same time, Uber gets a pledge that the car the team makes will be usable in the Uber network. They get guaranteed access to tech (if the new owner succeeds) though not the deep control that comes with owning it. They don’t need the deep control. In fact, it’s now better if they can stick another famous brand on it.

Nuro starts a delivery pilot

Aside from Starship (which I am involved in) the company to watch in the delivery space is Nuro. Nuro has started a delivery pilot in Scottsdale with Fry’s groceries. The pilot will be done with regular Prius and Leaf cars with Nuro’s self drive tech, and a safety driver, not with the small no-seat delivery pod Nuro is trying to build. That pod isn’t ready for the real streets, even the easy streets of Scottsdale.

Nuro is the company to watch in on-road delivery because its founders were among the top members of Google/Waymo’s team. I worked with them regularly when I was there and they are the best. They have also raised lots of money.

Driving on the road is much harder than the sidewalk, though. A road robot must be able to handle any situation on its own and get to a safe state if there is a problem no matter what. A sidewalk robot can stop in 30cm. If it encounters something it doesn’t understand, it can just stop and request help from HQ. Safety is very unlikely to be an issue, while a road robot could really hurt somebody in a mistake.

The $6 price for delivery in the pilot is strange, since that’s a common price for human driven delivery.

By Adam Conner-Simons | Rachel Gordon

Investigating inside the human body often requires cutting open a patient or swalloing long tubes with built-in cameras. But what if physicians could get a better glimpse in a less expensive, invasive, and time-consuming manner?

A team from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) led by Professor Dina Katabi is working on doing exactly that with an “in-body GPS” system dubbed ReMix. The new method can pinpoint the location of ingestible implants inside the body using low-power wireless signals. These implants could be used as tiny tracking devices on shifting tumors to help monitor their slight movements.

In animal tests, the team demonstrated that they can track the implants with centimeter-level accuracy. The team says that, one day, similar implants could be used to deliver drugs to specific regions in the body.

ReMix was developed in collaboration with researchers from Massachusetts General Hospital (MGH). The team describes the system in a paper that’s being presented at this week’s Association for Computing Machinery’s Special Interest Group on Data Communications (SIGCOMM) conference in Budapest, Hungary.

Tracking inside the body

To test ReMix, Katabi’s group first implanted a small marker in animal tissues. To track its movement, the researchers used a wireless device that reflects radio signals off the patient. This was based on a wireless technology that the researchers previously demonstrated to detect heart rate, breathing, and movement. A special algorithm then uses that signal to pinpoint the exact location of the marker.

Interestingly, the marker inside the body does not need to transmit any wireless signal. It simply reflects the signal transmitted by the wireless device outside the body. Therefore, it doesn’t need a battery or any other external source of energy.

A key challenge in using wireless signals in this way is the many competing reflections that bounce off a person’s body. In fact, the signals that reflect off a person’s skin are actually 100 million times more powerful than the signals of the metal marker itself.

To overcome this, the team designed an approach that essentially separates the interfering skin signals from the ones they’re trying to measure. They did this using a small semiconductor device, called a “diode,” that mixes signals together so the team can then filter out the skin-related signals. For example, if the skin reflects at frequencies of F1 and F2, the diode creates new combinations of those frequencies, such as F1-F2 and F1+F2. When all of the signals reflect back to the system, the system only picks up the combined frequencies, filtering out the original frequencies that came from the patient’s skin.

One potential application for ReMix is in proton therapy, a type of cancer treatment that involves bombarding tumors with beams of magnet-controlled protons. The approach allows doctors to prescribe higher doses of radiation, but requires a very high degree of precision, which means that it’s usually limited to only certain cancers.

Its success hinges on something that’s actually quite unreliable: a tumor staying exactly where it is during the radiation process. If a tumor moves, then healthy areas could be exposed to the radiation. But with a small marker like ReMix’s, doctors could better determine the location of a tumor in real-time and either pause the treatment or steer the beam into the right position. (To be clear, ReMix is not yet accurate enough to be used in clinical settings. Katabi says a margin of error closer to a couple of millimeters would be necessary for actual implementation.)

“The ability to continuously sense inside the human body has largely been a distant dream,” says Romit Roy Choudhury, a professor of electrical engineering and computer science at the University of Illinois, who was not involved in the research. “One of the roadblocks has been wireless communication to a device and its continuous localization. ReMix makes a leap in this direction by showing that the wireless component of implantable devices may no longer be the bottleneck.”

Looking ahead

There are still many ongoing challenges for improving ReMix. The team next hopes to combine the wireless data with medical data, such as that from magnetic resonance imaging (MRI) scans, to further improve the system’s accuracy. In addition, the team will continue to reassess the algorithm and the various tradeoffs needed to account for the complexity of different bodies.

“We want a model that’s technically feasible, while still complex enough to accurately represent the human body,” says MIT PhD student Deepak Vasisht, lead author on the new paper. “If we want to use this technology on actual cancer patients one day, it will have to come from better modeling a person’s physical structure.”

The researchers say that such systems could help enable more widespread adoption of proton therapy centers. Today, there are only about 100 centers globally.

“One reason that [proton therapy] is so expensive is because of the cost of installing the hardware,” Vasisht says. “If these systems can encourage more applications of the technology, there will be more demand, which will mean more therapy centers, and lower prices for patients.”

Katabi and Vasisht co-wrote the paper with MIT PhD student Guo Zhang, University of Waterloo professor Omid Abari, MGH physicist Hsaio-Ming Lu, and MGH technical director Jacob Flanz.

Sony on Thursday announced that its Aibo robotic dogs infused with artificial intelligence will be unleashed on the US market by the year-end holiday season, with a price tag of $2,899.