Computing Community Consortium Blog

The goal of the Computing Community Consortium (CCC) is to catalyze the computing research community to debate longer range, more audacious research challenges; to build consensus around research visions; to evolve the most promising visions toward clearly defined initiatives; and to work with the funding organizations to move challenges and visions toward funding initiatives. The purpose of this blog is to provide a more immediate, online mechanism for dissemination of visioning concepts and community discussion/debate about them.

Robots and Disasters: The Last 10 Years, The Next 2 Years

April 11th, 2011 / in policy, research horizons / by Ran Libeskind-Hadas

I’ve just returned from Japan advising the “Mt. Fuji” team on UAVs for the Fukushima nuclear situation and I’ll be going back next week with a team of experts and robots to assist several prefectures with tsunami damage inspection and the grim task of underwater victim recovery.  (Read more about this in The New York Times.)  This will be the Center for Robot-Assisted Search and Rescue’s (CRASAR) twelfth response since the first use of rescue robots at the 9/11 World Trade Center collapse just short of a decade ago.

Active Scope Camera (image courtesy International Rescue Systems Institute)What has changed in rescue robotics in the last 10 years?  The robots, of course!  Rescue robots originally meant small ground vehicles able to penetrate deep in the rubble.  Research advances have created increasingly smaller and more agile platforms and sensors, most strikingly the Active Scope Camera — an 18 foot long caterpillar-looking device (and indeed it moves like a hairy caterpillar, not a snake, which is astounding) developed by the International Rescue Systems Institute in Japan and used in partnership with CRASAR at the 2007 Berkman Plaza II Collapse.

But rescue robots are no longer just ground vehicles.  They include small aerial vehicles (which we introduced at Hurricane Katrina) and marine vehicles (another CRASAR first, at Hurricane Wilma).  The major criteria are that a robot can do things that people or dogs cannot (flying and diving count); can be transported as luggage on a plane and then be carried in a backpack or by a couple of people into a disaster zone; and can enable the responders to see what the robot is seeing in real time.

Another change in rescue robotics over the past 10 years is the human-robot interaction.  The World Trade Center deployments showed that the major barriers were in human comprehension:  the robots were agile enough but the interfaces and overall conceptualization of how to use the robots didn’t support the cognitive processes of the operators.  We literally couldn’t see what was right in front of us.  The analysis started a global re-examination of assumptions about the human-robot ratio and what type of intelligence and interfaces it would take to have one person control multiple robots.  There have been 10 years of cross-fertilization between deployments, experiments, and numerous investigations by several researchers of similar “remote assessment” domains such as bomb squads, IED robots, and wilderness search and rescue.  And at the end of the 10 years, the HRI research community has confirmed that perceiving and acting through computer mediation is very hard and leads to subtle, but real, misinterpretations of a remote situation.  Furthermore, as network communications have improved for disaster management, we’ve found that the real question isn’t how to have one person control multiple robots at a disaster, but rather how to enable that 10 or 100 people simultaneously view and direct with a single robot.  (Think about how many different engineers and decision makers need to know what is going on in the reactor buildings at Fukushima and how they can use the same set of imagery in different ways.)

But what is (painfully) striking is what hasn’t changed.  There are, practically speaking, no more rescue robots in use now than 10 years ago.  And there are still no direct mechanisms to get robots engaged in a disaster.  Only one search and rescue team in the U.S. owns ground rescue robots:  the New Jersey Task Force 1, which used these robots to go into the rubble of a recent parking garage collapse in Hackensack.  FEMA teams won’t be able to purchase rescue robots until the initial set of standards are completed by NIST, expected next year, but there’s no money set aside for purchases.  Many groups assume the small bomb squad/IED robots seen in the Hurt Locker will work for search and rescue but in reality they are too large and too specialized to be of much use.  The U.S. Military has micro fixed wing and rotary aircraft that are very appealing for disasters, but most of these are deployed overseas, and it is more difficult than you can imagine to borrow them.  And marine concepts of operation for disasters are very new, despite the relative maturity of those platforms, so only a few platforms are owned by law enforcement not fire rescue departments.

Sadly, the cache of robots maintained by CRASAR on constant readiness plus what we can tap from industry through our humanitarian Roboticists Without Borders program is it.  Our cache remains the core of proven, fieldable rescue robot — but that cache is too small, some of the robots are outdated, and all are too far away from international disasters.

What we need is a national program that creates a “virtual pool” of rescue robot technology where the latest land, sea, and air technology can be matched for the particular needs of a disaster and then rapidly transported on-site with appropriate logistical support.  Imagine a volunteer registry program, in which companies and universities can list their available robots with the results of the NIST ASTM testing (if applicable, as the standards don’t cover robots such as the Active Scope Camera).  And imagine that, if deployed, these robots had to collect performance data which would be shared and could help direct university research and corporate R&D, and better inform acquisition officers as to what is working under what circumstances.

But that’s not enough — as having a potentially useful technology isn’t the same as having the requisite field experience to actually fit in and use technology during a disaster.  So imagine the registrants getting funding for training (Disaster 101) and realistic field exercises with actual responders — which CRASAR has pioneered and hosted throughout the country, including the NASA Ames Disaster Assistance Response Team facility, and especially at our home site — Texas A&M’s Disaster City.

But having a registry of robots with trained, savvy operators isn’t enough either, because the robot teams have to be invited by the agency managing the incident.  You don’t just show up at a disaster; there are all sorts of reasons, including the liability of the incident commander, why you have to have an formal invitation.  (I know of a case where research scientists from a university showed up and started using their sensor gear at a disaster without permission and were actually jailed overnight for trespass and obstruction.  And no, this didn’t have a “this is all a misunderstanding, we’re sorry, and we could we use your wonderful gear” ending the next morning.  Incident commanders are very busy and under a lot of stress, and if you violate the rules, well, jail is one possible outcome.)

Transportation is a problem as the first 72 hours are key in finding survivors who will recover without long-term disabilities.  Our cache can be ready to go in four hours.  Yet the earliest we could have gotten to Japan would have been 36 hours due to airline schedules, assuming we got an invitation within 18 hours of the earthquake (which we didn’t; requests started coming in after two days but by then the escalating nuclear situation prevented us from traveling).  So without private or military travel, the value of rescue robots for search and rescue (versus recovery operations) depends on the luck of the draw on commercial airlines.  And even domestic flights on short notice are expensive.

Therefore, we need a liaison for U.S. disasters (no, it wouldn’t necessarily be DHS, as mine and energy disasters don’t fall under their purview, another added bonus complication of emergency response) and State Department and Defense Department liaisons for international disasters to help arrange the invitations, transportation, sharing of military resources, and logistics.

But wait — there’s more.  Even with registries, training, liaisons, there’s still one missing ingredient:  some form of compensation.  Companies, especially startups, and universities can work only for so long without pay.  But disasters take years to clean up.  After the initial life-saving phase, there’s the recovery phase — the inspection of bridges, ports, and other infrastructure that I mentioned earlier.  (And in the Great East Japan Earthquake case, victim recovery as well.)  So what if the equipment and personnel costs for the first 10 days were subsidized directly by the U.S., and then if an agency or country liked what it saw, there was an easy way to pay for those services, or subsets, to continue?  There has to be a way to create a workable, equitable solution.

Wouldn’t it be a great change for the next 10 years of rescue robotics to actually use the robots that our country’s research investment has produced?

And why wait 10 years, why not make it 2 years?

(Contributed by Robin Murphy, Texas A&M University)

Robots and Disasters:  The Last 10 Years, The Next 2 Years