The White House Office of Science and Technology Policy recently organized a symposium of top robotics experts at Worcester Polytechnic Institute to brainstorm how field robots could be used in future Ebola-like pandemics. While the researchers came up with a number of innovative short-term and long-term ideas for how robots could be used to fight Ebola – everything from cleaning and decontaminating rooms to actually administering IVs to humans under medical treatment — there are still a number of important issues to clarify before we hand over the task of fighting Ebola to the robots.
On the surface, of course, handing over the dirty work of cleaning up after an Ebola outbreak to the robots sounds like a no-brainer. Instead of putting humans into harm’s way, why not just send in a robot? Robots can’t develop symptoms from Ebola, they are relatively easy to disinfect (except for their wheels), they dutifully carry out tasks without talking back and they can dispose of hazardous waste efficiently.
Scratch the surface, though, and you can start to see the moral and philosophical questions that arise once robots start doing more than just grunt-level decontamination work. In short, everything changes once robots also become human-like caregivers of Ebola patients rather than just repurposed industrial robots. Even assuming that wise and highly moral technologists have created robots according to something approximating Isaac Asimov’s Three Laws of Robotics, there still exists all kinds of potential for things to go wrong as robots go about trying to observe these laws. Just read any of Asimov’s “Robot” stories (or, better yet, watch the Will Smith movie) to understand how things might go awry.
TEEX trainer in protective gear with a “MUTT” robot carrying a litter at the workhop demo.
CRASAR, with funding from the Center for Emergency Informatics, and the TEEX Product Development center held a two day series of workshops on robotics for medical disasters. The major takeaway was that robots do exist that could be immediately repurposed now to protect Ebola health workers but how robots fit into the medical response enterpriseis as important as what the robots can actually do. While most roboticists intuitively know that what will work in the US is not the same as what will work in West Africa, the differences go beyond physical constraints such as level floors, ample power, and reliable wireless communications infrastructure. Less intuitive is that the cultural appropriateness of the technology and the impact on the existing workflows and practices is equally important.
The workshops considered how robots could be used immediately and in future domestic medical responses. Hardened robots (and automation/CPS technologies) do exist that could be immediately repurposed to provide logistical services (e.g., packing and hauling contaminated waste) and reconnaissance (e.g. observing signs of mass graves near a village), less so for clinical applications (e.g., directly working with patients). The participants strongly concurred that a research roadmap is needed to prepare robots that the US can effectively use in future medical disasters.
The success of hardened robots in providing these services depends on ensuring that they are appropriate for the work domain in five ways:
Fit the cultural context. For example, a telepresence robot allowing a certified medical interpreter to talk with the family and talk the patient’s history may overwhelm a non-Western family who has never seen a computer. A less obtrusive telepresence solution may be more practical in that cultural context.
Fit the existing workflow and practices. For the short term, solutions aren’t solutions if they require health workers or medical responders to adopt radically new procedures. They simply can’t handle more things to do or change how they perform their current tasks (which impacts how everyone performs all the other tasks- “simple” changes can have system ramifications). However, small changes that produce at least a tenfold benefit can make a difference.
Can function in the target environment. For example robots in West Africa would have work reliably in field hospitals with canvas floors and narrow doors, muddy dirt roads in the rainy season, with power and wireless communications limitations, etc., while robots in the US would have more pristine conditions. Different groups use different decontamination procedures and chemicals- such as dousing everything with chlorine beach solution (easy and inexpensive) or using more chemically sophisticated decontamination foams used by urban hazardous materials teams.
Are maintainable and sustainable. Health workers and medical responders won’t have the time and skills to repair robots (especially if wearing PPE) and may not have access to consumables such as batteries to enable operations for weeks and months. A problem with the Fukushima response was that many robots were actually prototypes functioning at a Technical Readiness Level of 7 rather than a well-tested Level 9 system.
Are easy to use and be trained on. This is related to fitting into the existing workflow and practices, but deserves special emphasis. The health workers and responders will not have significant amounts of time to learn new tools, as their days are already overloaded and they have little personal time. Robots must be vetted for ease of use. Effective training for medical missions is important and the role of simulation or serious games should not under-estimated.
The sentiment shared by the TAMU participants was that the biggest barrier to near-term use was not the lack of capable robots but rather the lack of requirements that would allow industry to invest in repurposing robots and enable agencies to test and evaluate the robots and develop training.Currently there are no details on the operational envelopment for the robot or operator. There is no clearinghouse of social science data on cultural appropriateness or bioethics or specific missions.
OVERVIEW OF WORKSHOP ACTIVITIES
The first day of the workshop was hosted by CRASAR and held at the National Center for Therapeutic Manufacturing. The day was divided into two portions. One was a simulcast of shared presentations with the other three sites and brainstorming as part of the planning workshops on Safety Robotics for Ebola Workers for the White House OSTP/National Robotic Initiative. The other part, the Texas A&M Workshop on Robotic, Automation and Cyber Physical Systems for MedicalResponse to Disasters, provided additional talks and discussions on general domestic medical response. The Texas A&M talks covered the state of the practice in DoD robots (TARDEC) and casualty evacuation systems (TATRC) that can be repurposed, lessons learned so far in using robots at the Fukushima Daiichi decommissioning (University of Tokyo), and opportunities for community recovery (TAMU Hazards Reduction and Recovery Center). The day culminated with a reception and a thought provoking keynote talk by Andrew Natsios (TAMU Bush School of Government and Public Policy), who served as administrator of USAID from 2001-2006.
The second day, the Infectious Disease Response Workshop, was hosted by Caleb Holt and the TEEX Product Development Center and held at the TEEX Emergency Operations Training Center/Disaster City® complex. The focus was on the practice of medical response (one of TEEX many courses that they teach). A major portion of the day was spent in demonstrations of the current practices in medical response, walking participants through 3 modules of a field hospital (also called an Emergency Treatment Unit or ETU), showing how contaminated waste is stored and overpacked, and how domestic responders, equipment, and ambulances are decontaminated. One demonstration was not a current practice but showed how existing robots might be of use. That demonstration showed the General Dynamics Land Systems MUTT, a robot wagon that acts like a dog and can carry waste, supplies, or one or two litters. A responder guides the robot with a leash rather than a video game controller that is hard to carry and use while wearing personal protective equipment. If the responder stops, the robot stops. If the person backs up, the robot backs up. The second day also featured panels of practitioners, including from the Texas Ebola Task Force and the USMC Chemical Biological Incident Response Force, comparing military and domestic practices
TAMU FINDINGS ON WHAT ROBOTS CAN BE USED FOR
In terms of overall medical disasters, applications appear to fall into one of three broad categories below, regrouping the preliminary list of nine functions discussed in an earlier blog. Each category has a different set of stakeholders and a different operational envelope that the robots would operate in. Clinical applications are possibly what people think of first– how robots can replace what health workers do now—but logistical applications are perhaps the most feasible and practical.
Clinical: Clinical applications are where robots are used in the ETU as a “force multiplier” (another way of saying “reducing manpower”) by taking over some of the activities that health workers do or as adding reliability by coaching or supervising activities. Ignoring for a moment the cultural appropriateness and other adoption issues, robots could enable
Remote health workers to assist other health workers, such as telepresence robots (or just cameras/tablets) coaching or supervising taking off PPE– though the general consensus of our responder base was that having a second person physically helping with decon was more valuable than having someone saying “hey, you touched your face while trying to lift your hood.” Domestic hazmat responders and the USMC Chemical Biological Incident Response Forces use a two personal decontamination process.
Health workers could use robots to interact with patients, reducing the number of times workers have to risk exposure. Robots could provide non-invasive point-of-care such as changing IV bags, though the TAMU participants were more reserved about roboticizing invasive procedures such as starting IV lines.
Remote health workers to interact with family members, such as remote qualified medical interpreters working through telepresence to help with patient intake forms.
Logistical: Logistical applications can take place within the ETU, but the construction, layout, and clutter of ETUs make it hard for mobile robots to move around. Some ETUs have canvas floors over dirt or mostly level manufactured floors, and almost all have raised areas to step over between modules that seem intended to foul wheels. The general thought is that flexible automation and materials handling are more likely to be of benefit within an ETU and that robots would be more useful for outside the ETU. Logistical robots are also interesting in terms of stakeholders. Since they are not performing clinical functions, in theory the robots could be operated by locals (assuming favorable cultural considerations).
Logistical robots could provide
Materials handling. Robots could reduce the number of times humans handle contaminated waste or the number of people needed to carry a litter. The robots could pack and carry materials from the warm zone to the cold zone (e.g., taking out the trash) or carry supplies into the warm zone, saving another cycle of a person having to don and doff PPE.
Decontamination. Robots could spray biocide foam on equipment, though there was several ideas for using gases to rapidly decontaminate ambulances so as to keep them in service.
Delivery and resupply. Unmanned aerial vehicles or boats could drop off small amounts of supplies to villages cut off by the rainy season.
Reconnaissance: Reconnaissance activities take place outside of the ETU. Aerostats or UAVs can provide awareness of long lines or gathering refugees. A more somber recon activity is to fly over villages and look for signs of freshly turned earth indicating graves.
Other: The workshops also touched on preparation for medical response, such as redesigning field hospitals to make it easy to use robots and to add cameras, internet repeaters, etc. The workshops raised the value of automated construction in reducing the non-medical members of the team needed to set up and maintain the ETU.
The Netherlands has unveiled a prototype of an “ambulance drone” which could potentially save lives by offering a rapid response after heart attack incidents. The yellow-painted drone can gain speeds up to 100 km per hour and is equipped with a defibrillator. The unmanned aerial vehicle (UAV) is flown with the aid of six propellers and can carry up to four kilograms, according to the creator of the project, Alec Momont, who revealed the first test model at the Netherlands’ Delft University of Technology.
A new robotic tool could help rescue workers locate victims of disasters and other emergencies before venturing into collapsed buildings or other potentially dangerous places.
Developed by researchers at the University of Guadalajara in Mexico, the new robotic system enables small, rugged bots — designed for search-and-rescue missions — to distinguish between human bodies and other objects, such as piles and rubble.
As the robot roams around a disaster site, it snaps pictures of its surroundings with the 3D camera and then sends those images to the computer. The computer then scans the images for patterns that might indicate the presence of a human body, using a specially created algorithm. The algorithm must first break down visual information into mathematical data by using what’s known as a descriptor system, which assigns numerical values to different parts of the 3D images. The numbers represent the different shapes, colors and densities of the objects in the picture.
All of this mathematical data is then merged together to create a second, much simpler, image. This image is passed through another algorithm, which detects whether the object that appears in the new image is a person or something else.
The difference between life and death for a troubled soul who jumps over the Sunshine Skyway Bridge, which is ranked fourth for suicide jumps in the United States, can rest in the hands of a group of students on the Eckerd College Search and Rescue Team. There are about 50 students and three paid staff members on the team.
While the 24-7 team is the only volunteer, college-based marine search and rescue group in the country, there are other unique teams around the nation. Texas A&M has a Center for Robot Assisted Search and Rescue in College Station and a handful of schools perform backcountry searches.
Texas A&M is one of the four sites co-hosting a OSTP/NRI Workshop on Safety Robotics for Ebola Workers. Our workshop will be November 7-8, with November 7 coordinated with the other three sites and November 8 as a follow-on at Disaster City specifically on technology transfer. We are still working on the agenda, but attendance is limited and by invitation. Participants need to be physically at College Station in order to help generate and rank the list of opportunities for robotics to give to the White House and to work with the medical and humanitarian responders to elicit operational details critical for successful technology transfer. Attached are some photos of a 2004 robotics exercise we hosted with the USMC Chemical Biological Incident Response Force- as you can see we learned a lot about working with PPE. Likewise our involvement in the Fukushima Daiichi nuclear accident reinforced and amplified how little things can trip up responses.
Our site’s workshop will address how robots can be used beyond protecting Ebola workers and that it will focus not only on helping identify what robots can do but on how robots must do it in order to be successful. Here at A&M we are striving to create a set of detailed use cases and projected robot requirements that can be used by industry and the TEEX Product Development Center. The robotics community cannot provide robots without understanding the needs otherwise engineering mistakes or mismatches that will be both financially costly and delay the delivery of effective solutions.
Robot carrying a victim at CRASAR/USMC CBIRF exercise (2004)
To meet these objectives, our workshop is focused on working with medical and humanitarian relief experts (they talk, we listen) to answer two questions:
what are the most pressing problems, barriers, or bottlenecks? e.g. minimizing contact while burying bodies or disposing of waste, health worker protection from infection,decontamination and disinfection of facilities, detection of presence of Ebola in facilities,tele-consulting by remote experts, health work physical safety, delivery of supplies, etc.
What is the value proposition of using a robot? e.g., benefits versus manpower, logistics support, training requirements, economic costs, etc.Is a robot the best choice? For example, Dr. Mark Lawley here in Industrial and Systems Engineering is working on adapting low-cost flexible manufacturing methods for waste and materials handling within the field hospitals where a mobile robot would be a technological overkill.
In my previous blog, I described 9 categories of applications that we’ve identified so far for robots for Ebola. It’s fairly easy to come up with ideas and there is a wealth of ground, aerial, and marine robots that can be repurposed. But it’s much harder to determine what’s the real value to the medical and humanitarian responders and to ferret out those hidden requirements that support a successful technology transfer. Our research and hands-on experiences at CRASAR has shown that military robots have not been a perfect match for fire rescue and law enforcement and many attempts by vendors to deploy their robots to disasters or to sell their robots to the homeland security community have failed. I see these failures stemming from three types of constraints: the operational envelope, work domain,and culture.
The operational envelope focuses on workspace attributes such as environmental conditions, size of doors in field hospitals, communication and power infrastructure, etc. As detailed in Disaster Robotics, several types of rescue robots were not used at the 9/11 World Trade Center because they could not fit in the luggage bays on buses hired to carry FEMA search and rescue teams. Some concerns about robots such as how can robots be decontaminated become moot if the robot can be recharged and maintained by workers inside the Hot Zone so that it never needs decon– but this of course means that functions can be performed by workers wearing personal protection equipment.
The work domain is critical as anyone who works in system design knows. Who are the stakeholders? Will these robotic solution employ locals so as to help support their economy? If so, what does that mean in terms of making robots that are easy to use and reliable? We use a formal method called cognitive work analysis to determine the work domain.
Culture is technically part of the work domain, but I personally think it merits special attention. We robot designers need to have cultural sensitivity to customsfor caring for the ill and conducting burials if we create robots to tend to the sick and transport the deceased. The rhythms of village life also impact humanitarian relief, for example it is better for a medium sized UAV to drop off a large payload of supplies and let the village equivalent of the American Red Cross representative go fetch it and deliver it to different households as part of their daily routine or should a smaller UAV do a precision drop to individuals?
Robot operator’s view from the controller
Robot eye view of victim being transported so medical person can make sure they aren’t having a seizure, etc.
As fears continue to grow over the recent outbreak of Ebola, scientists and researchers in the U.S. are hoping to develop a strategy for combating the virus’ spread through the use of robots and autonomous vehicles. November 7th will see workshops put together by the Center for Robot-Assisted Search and Rescue that brings robotocists together with members of the medical and humanitarian aid communities to hopefully find a solution.
The initial idea is that depending on the situation, robots can be used as mobile interpreters, methods of delivery for much-needed supplies such as medicine and food, and even during the most dangerous of tasks like decontamination or burying deceased victims. “What are the things robotics can do to help?” poses Robin Murphy, a robotics professor at Texas A&M University, as well as the director of the Center for Robot-Assisted Search and Rescue. One idea put forward by a robotics engineer is to take a wheeled robot and attach two decontamination sprayers, and then have it work in places where the virus has been found, or on cleaning equipment.
What is being stressed leading up to the workshops is that robots are not act as full replacements for human aid workers. The goal is to minimize workers’ contact, but for every piece of technology put to use, there should still be a human to interact with.
There have been many nature-inspired gadgets and devices. This method is called biomimicry or biomimemtics. You have products like shark skin which inspired swim suits and submarine coatings; Velcro, the hook and loop fastener that was inspired by plant burrs that stick to dog hair and a new adhesive inspired by Geckos. Now, snakes have inspired the creation of a robot that will mimic its actions and can move through tiny holes.
It is said that the snake robot has been in use since 2008, but these were early prototypes, since which many alterations have been made in different models. The most famous is Carnegie Mellon University’s (CMU) research and snake bot models. Georgia Tech University students have also developed search and rescue snake robots in 2012.
Recently researchers from Carnegie Mellon and Georgia Tech headed to Zoo Atlanta, to observe rattlesnakes. After over 50 trials, these snake movements were measured and tracked through cameras. “The snakes tended to increase the amount of body contact with the surface at any instant in time when they were sidewinding up the slope and the incline angle increased,” said Daniel Goldman, co-author of the study and an associate professor of biomechanics at the Georgia Institute of Technology in Atlanta.
These researchers also got in touch with Howie Choset, a professor at The Robotics Institute at CMU. Prof Howie has been working on developing limbless snake bots that can move through small spaces. The professor said these robotic sidewinding abilities could come in handy in archaeological sites. For instance, the robots could be used to explore the insides of pyramids or tombs. They could also help in search and rescue missions, as they are capable of moving in small and cramped spaces.
Highvale’s Scott Nicholson and Albany Creek’s Jaynesh Vanmali are part of a Queensland University of Technology team taking the Maritime RobotX Challenge.
The competition is designed to increase the autonomy of robotic boats so they could perform real-world tasks in real-world environments, including searching for debris or oil slicks or finding overboard mariners in rough seas.
Running from 20-26 October, the Maritime RobotX Challenge involves 14 teams from universities in Australia, Singapore, Japan, USA and South Korea.
“Teams are judged on how competent their boat is at completing tasks,” Mr Vanmali said. “There are five tasks in total and each of them assess the boat’s ability in docking, navigation, obstacle avoidance and search and rescue.”
I’ve been working since Sept 17 on robots for the Ebola epidemic– both in terms of what can be used now and what can be used for future epidemics. Dr. Taskin Padir at WPI deserves a big shout out for calling the robotics community’s attention to this, with Gill Pratt at DARPA and head of the DARPA Robotics Challenge and Richard Voyles Associate Dean at Purdue.
I am pleased to announce that CRASAR will be co-hosting a White House Office of Science and Technology Policy workshop on Safety Robotics for Ebola Workers on Nov. 7. Texas A&M was already planning a medical response workshop on the 7th for disasters in general, so expanding that to a virtual event over the internet with sessions at the White House (OSTP and DARPA), Boston (Taskin), and Berkeley (Ken Goldberg). CRASAR is already planning to host another workshop to share the results of our current research into specific use cases with the robotics community in the Jan 3-15, 2015, timeframe.
Here on campus, students will be creating prototypes as part of the Aggies Invent event What Would You Build for a First Responder event on Oct. 24-27 and the students in my graduate AI Robotics class this semester will be designing and simulating intelligent robots.
The real issue to me is what are the real needs that robots can play in such a complex event? Here are some possibilities that have emerged in discussions and I am sure that there are many more (let me know what you think!):
Mortuary robots to respectfully transport the deceased, as ebola is most virulent at the time of death and immediately following death
Reducing the number of health professionals within the biosafety labs and field hospitals (e.g., automated materials handling, tele robotics patient care)
Detection of contamination (e.g., does this hospital room, ambulance or house have ebola)
Disinfection (e.g., robots that can open the drawers and doors for the commercially available “little Moe” disinfectant robot)
Telepresence robots for experts to consult/advise on medical issues, train and supervise worker decontamination to catch accidental self-contamination, and serve as “rolling interpreters” for the different languages and dialects
Physical security for the workers (e.g., the food riots in Sierre Leone)
Waste handling (e.g., where are all the biowaste from patients and worker suits going and how is it getting there?)
Humanitarian relief (e.g., autonomous food trucks, UAVs that can drop off food, water, medicine, but also “regular” medicine for diabetes, etc., for people who are healthy but cut off)
Reconnaissance (e.g., what’s happening in this village? Any signs of illness? Are people fleeing?)
In order to be successful at any one of the tasks, robots have to meet a lot of hidden requirements and sometimes the least exciting or glamorous job can be of the most help to the workers. Example hidden requirements: Can an isolated field hospital handle a heavy robot in the muddy rainy season? How will the robots be transported there? Is it easy enough for the locals to use so that they can be engaged and earn a living wage? What kind of network communication is available? What if it needs repairs? That’s what I am working on, applying the lessons learned in robotics for meteorological and geological disasters.
I am certainly not working alone and am reaching out to experts all over the world. In particular, four groups have immediately risen to the challenge and are helping. Matt Minson MD and head of Texas Task Force 1’s medical team and Eric Rasmussen MD FACP (a retired Navy doctor) who has served as the medical director for the Center for Robot-Assisted Search and Rescue since 9/11 have offered their unique insights. There are two DoD groups: the USMC Chemical Biological Incident Response Force (the team that cleaned up the anthrax in DC) with whom I’ve served on their technical advisory board and the Army Telemedicine & Advanced Technologies Research Center (TATRC), where Gary Gilbert MD has led highly innovative work in telemedicine and in casualty evacuation (Matt and I had a grant evaluating robotic concepts).
