In this episode, Audrow Nash interviews Pauline Pound, Philippe Morere, and Yujung Liu about the work they presented at the 2018 International Conference on Intelligent Robots and Systems (IROS) in Madrid, Spain.
Pauline Pounds, Associate Professor at the University of Queensland, speaks about building robots that can endure children. She discusses the tradeoffs of designing a robot that can survive children and cannot harm the children. Pounds talks about how the robot has performed with children so far, the hardware design, and her future direction with this work.
Philippe Morere, a PhD student at University of Sydney, speaks about reinforcement learning in partially observable environments. Morere discusses the intuition behind this work and the problems that he solved using this approach.
Yujung Liu, a master’s student from the National Taiwan University, speaks about a painting robot. He discusses how the portrait is constructed, including how colors are chosen, how the features are made cartoon-like, and his future direction.
Links
By Leah Burrows
Children born prematurely often develop neuromotor and cognitive developmental disabilities. The best way to reduce the impacts of those disabilities is to catch them early through a series of cognitive and motor tests. But accurately measuring and recording the motor functions of small children is tricky. As any parent will tell you, toddlers tend to dislike wearing bulky devices on their hands and have a predilection for ingesting things they shouldn’t.
Harvard University researchers have developed a soft, non-toxic wearable sensor that unobtrusively attaches to the hand and measures the force of a grasp and the motion of the hand and fingers.
The research was published in Advanced Functional Materials and is a collaboration between the Wyss Institute for Biologically Inspired Engineering, The Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS), Beth Israel Deaconess Medical Center, and Boston Children’s Hospital.
One novel element of the sensor is a non-toxic, highly conductive liquid solution.
“We have developed a new type of conductive liquid that is no more dangerous than a small drop of salt water,” said Siyi Xu, a graduate student at SEAS and first author of the paper. “It is four times more conductive than previous biocompatible solutions, leading to cleaner, less noisy data.”
The sensing solution is made from potassium iodide, which is a common dietary supplement, and glycerol, which is a common food additive. After a short mixing period, the glycerol breaks the crystal structure of potassium iodide and forms potassium cations (K+) and iodide ions (I-), making the liquid conductive. Because glycerol has a lower evaporation rate than water, and the potassium iodide is highly soluble, the liquid is both stable across a range of temperatures and humidity levels, and highly conductive.
“Previous biocompatible soft sensors have been made using sodium chloride-glycerol solutions but these solutions have low conductivities, which makes the sensor data very noisy, and it also takes about 10 hours to prepare,” said Xu. “We’ve shortened that down to about 20 minutes and get very clean data.”
The design of the sensors also takes the need of children into account. Rather than a bulky glove, the silicon-rubber sensor sits on top of the finger and on the finger pad.
“We often see that children who are born early or who have been diagnosed with early developmental disorders have highly sensitive skin,” said Eugene Goldfield, Ph.D., coauthor of the study, and Associate Faculty Member of the Wyss Institute at Harvard University and an Associate Professor in the Program in Behavioral Sciences at Boston Children’s Hospital and Harvard Medical School. “By sticking to the top of the finger, this device gives accurate information while getting around the sensitively of the child’s hand.”
Goldfield is the Principal investigator of the Flexible Electronics for Toddlers project at the Wyss Institute, which designs modular robotic systems for toddlers born prematurely and at risk for cerebral palsy.
Goldfield and his colleagues currently study motor function using the Motion Capture Lab at SEAS and Wyss. While motion capture can tell a lot about movement, it cannot measure force, which it critical to diagnosing neuromotor and cognitive developmental disabilities.
“Early diagnosis is the name of the game when it comes to treating these developmental disabilities and this wearable sensor can give us a lot of advantages not currently available,” said Goldfield.
This paper only tested the device on adult hands. Next, the researchers plan to scale down the device and test it on the hands of children.
“The ability to quantify complex human motions gives us an unprecedented diagnostic tool,” says senior author Robert Wood, Ph.D., Founding Core Faculty Member of the Wyss Institute and the Charles River Professor of Engineering and Applied Sciences at SEAS. “The focus on the development of motor skills in toddlers presents unique challenges for how to integrate many sensors into a small, lightweight, and unobtrusive wearable device. These new sensors solve these challenges – and if we can create wearable sensors for such a demanding task, we believe that this will also open up applications in diagnostics, therapeutics, human-computer interfaces, and virtual reality.”
This research was also authored by SEAS researcher Daniel M. Vogt; Wyss Institute researchers Wen-Hao Hsu, John Osborne, Timothy Walsh, Jonathan R. Foster, Sarah K. Sullivan, and Andreas Rousing; and Beth Israel Deaconess Medical Center researcher Vincent C. Smith. It was supported by the National Institutes of Health.
PUBLICATION – Advanced Functional Materials : Biocompatible Soft Fluidic Strain and Force Sensors for Wearable Devices
WYSS TECHNOLOGY – “Flexi-mitts”for tracking neurodevelopment in very low birth weight premature infants
Ivar Mendez, University of Saskatchewan
It is the middle of the winter and a six-month-old child is brought with acute respiratory distress to a nursing station in a remote community in the Canadian North.
The nurse realizes that the child is seriously ill and contacts a pediatric intensivist located in a tertiary care centre 900 kilometres away. The intensivist uses her tablet to activate a remote presence robot installed in the nursing station and asks the robot to go to the assessment room.
The robot autonomously navigates the nursing station corridors and arrives at the assessment room two minutes later. With the help of the robot’s powerful cameras, the doctor “sees” the child and talks to the nurse and the parents to obtain the medical history. She uses the robot’s stethoscope to listen to the child’s chest, measures the child’s oxygen blood saturation with a pulse oximeter and performs an electrocardiogram.
With the robot’s telestrator (an electronic device which enables the user to write and draw freehand over a video image) she helps the nurse to start an intravenous line and commences therapy to treat the child’s life-threatening condition.
This is not science fiction. This remote presence technology is currently in use in Saskatchewan, Canada — to provide care to acutely ill children living in remote Northern communities.
Advances in telecommunication, robotics, medical sensor technology and artificial intelligence (AI) have opened the door for solutions to the challenge of delivering remote, real-time health care to underserviced rural and remote populations.
In Saskatchewan, we have established a remote medicine program that focuses on the care of the most vulnerable populations — such as acutely ill children, pregnant women and the elderly.
We have demonstrated that with this technology about 70 per cent of acutely ill children can be successfully treated in their own communities. In similar communities without this technology, all acutely ill children need to be transported to a tertiary care centre.
We have also shown that this technology prevents delays in diagnosis and treatment and results in substantial savings to the health-care system.
Remote communities often lack access to diagnostic ultrasonography services. This gap disproportionally affects Indigenous pregnant women in the Canadian North and results in increases in maternal and newborn morbidity and mortality.
We are pioneering the use of an innovative tele-robotic ultrasound system that allows an expert sonographer to perform a diagnostic ultrasound study, in real time, in a distant location.
Research shows that robotic ultrasonography is comparable to standard sonography and is accepted by most patients.
The first tele-robotic ultrasonography systems have been deployed to two northern Saskatchewan communities and are currently performing prenatal ultrasounds.
Portable remote presence devices that use available cellular networks could also be used in emergency situations, such as trauma assessment at the scene of an accident or transport of a victim to hospital.
For example, emergency physicians or trauma surgeons could perform real-time ultrasonography of the abdomen, thorax and heart in critically injured patients, identify life-threatening injuries and start life-saving treatment.
Wearable remote presence devices such a Google Glass technology are the next step in remote presence health care for underserviced populations.
For example, a local nurse and a specialist in a tertiary care centre thousand of kilometres away could assess together an acutely ill patient in an emergency room in a remote community through the nurse’s eyes.
Although remote presence technology may be applied initially to emergency situations in remote locations, its major impact may be in the delivery of primary health care. We can imagine the use of mobile remote presence devices by health professionals in a wide range of scenarios — from home-care visits to follow-up mental health sessions — in which access to medical expertise in real time would be just a computer click away.
The current model of centralized health care, where the patient has to go to a hospital or a clinic to receive urgent or elective medical care, is inefficient and costly. Patients have to wait many hours in emergency rooms. Hospitals run at overcapacity. Delays in diagnosis and treatment cause poor outcomes or even death.
Underserviced rural and remote communities and the most vulnerable populations such as children and the elderly are the most affected by this centralized model.
Remote presence technologies have the potential to shift this — so that we can deliver medical care to a patient anywhere. In this decentralized model, patients requiring urgent or elective medical care will be seen, diagnosed and treated in their own communities or homes and patients requiring hospitalization will be triaged without delay.
This technology could have important applications in low-resource settings. Cellular network signals around the globe and rapidly increasing bandwidth will provide the telecommunication platform for a wide range of mobile applications.
Low-cost, dedicated remote-presence devices will increase access to medical expertise for anybody living in a geographical area with a cellphone signal. This access will be especially beneficial to people in developing countries where medical expertise is insufficient or not available.
The future of medical care is not in building more or bigger hospitals but in harnessing the power of technology to monitor and reach patients wherever they are — to preserve life, ensure wellness and speed up diagnosis and treatment.
Ivar Mendez, Fred H. Wigmore Professor and Unified Head of the Department of Surgery, University of Saskatchewan
This article is republished from The Conversation under a Creative Commons license. Read the original article.
A number of other summaries of 2018 in robocars have called it a bad year, the year it all went south, even the year the public realized that robocars will never come.
In fact, 2018 was the year the field reached a new level of maturity, as its warts began to show, and we saw the first missteps (minor and major) and the long anticipated popping of some of the hype.
As predicted by Gartner’s famous “hype cycle” any highly-hyped technology goes through a “trough of disillusionment” after the initial surge.I see several reasons why the trough is happening now:
In reality, the plan at Waymo and many other companies has always been to build a car that serves a limited set of streets — a small set of easy streets — at the start, and then starts growing the network of streets, and eventually towns with streets, as time goes by.
There is going to be a big surge once the technology reaches a level where the remaining problems are no longer technological as logistic.That is to say, when the barrier to expanding to new streets and cities is the detailed work of mapping those streets, learning the local rules and working with local governments.That’s when the “land rush” happens.The limiting factor there is time and talent more than it’s money.
But none of that happens until the cars are ready for deployment, and until they are, they will be tested as prototypes with safety drivers in them.Even the first prototype services, like Waymo’s and Zoox’s and others, will have safety drivers in them.
No question the big story this year was the death of Elaine Herzberg as the result of a compound series of errors and bad practices at Uber.The story is notable for many reasons, including of course how it happened, but also in the public’s reaction. For a long time, I’ve been assured by many skeptics that the first death would mean the end of the robocar dream.The public actually thinks the first deaths were in Teslas (they weren’t) and Tesla stock went up after they took places. The Uber fatality was real, and did teach us that teams are capable of more negligence than I had thought. While it did scale up public distrust, and Uber did shut down their program for at least a year, the overall effect still seems modest.(The larger effect will be much greater intolerance for the next fatality, the one that would have been the first.)
Here’s some of my many posts on Uber this year:
Waymo remains the clear leader in the field, so the next top story has to be about them, but sadly it’s the story of their first miss — promising to launch in 2018 and feeling forced to do a “launch” that was really just a formalization of existing activity. I believe that Uber is partly to blame here, in that it did use up a lot of the public’s tolerance for errors, especially in the Phoenix area.Waymo soft launches in Phoenix, but
The better story for Waymo, however, was their first totally unmanned operations earlier in the year.This also disappointed people because these unmanned operations were on a much more limited scale than people originally imagined, but it’s still a major milestone.It means Waymo’s team convinced the lawyers and board that the systems were good enough to take this risk, even if only in a limited area.
Waymo goes totally unmanned, arbitration and other news
This was also the year that “flying cars” also known as e-VTOL aircraft, “took off.”It’s now clear the engineering problems are close to solved, though many social and logistic problems remain.These vehicles are at the stage robocars were 10 years ago, and the excitement is building.Sebastian Thrun, the modern “father of self-driving cars” and the man who first got me excited about them, has switched his efforts to flying.I’ll be writing more on this in the coming year.
In chronological order, not order of importance
The main focus of this site are my essays on the issues and future of robocars.Here are the ones from this year I think you will find most valuable.