Archive for the ‘Rescue Robots’ Category

About Unmanned Aerial Systems and the controversy over not using them at the Colorado Floods

The tone and content of the discussion of the lack of use of UAVs in the Colorado flooding among the community bothers me. I have flown small UAVs (or UAS depending on what agency is involved) at Hurricane Katrina and Wilma and have assisted with the UAV flights at Fukushima. I have also been turned down by the FAA for emergency COAs. I was involved in discussions to use UAVs at the Moore, Oklahoma, and the decision not to use them there.

So I feel I have unique perspective from both sides of the coin.

During a disaster, a UAS requires an emergency COA from the FAA, not FEMA. An agency or responsible party has to want to use the UAS, be it FEMA, a task force, or a sheriff before the FAA considers the request. If I as director of CRASAR ask the FAA, they would laugh at me; instead I make the request with the agency that CRASAR is working for (we don’t self-deploy which is illegal and unethical). The FAA can turn around an emergency COA in 30 minutes if the requestor can show that the UAS has an existing COA, can specify where it will fly, and show need.

Manned and unmanned aircraft are currently not allowed to operate over the same area at the same time, even though they may be at different altitudes. So if an agency wants to fly manned vehicles, such as CAP or manned helicopters trying to airlift people out, they have to decide if they want to stop those operations temporarily to let unmanned systems in. And remember, since communications with the air traffic controllers who are working the disaster and the actual pilots is spotty, there has to be a bit latency/down time to get the word out. So this could shut manned operations down in an area for quite a while.  The current solution is to partition the space, e.g., “this canyon on Thursday will be just for unmanned systems.”  Within that area and time frame, responders can use a small UAS on demand- though there’s always a provision to land immediately if an manned system enters the airspace for some reason.

However, having to plan to this level of detail, combined with the latency, will drive an agency such as FEMA to say it’s just not worth the time and effort- the cost-benefit to the _larger_ operations isn’t there. The cost of the deployment has never been an issue, free or not- it just may not be useful.  That’s what happened at Moore. There were manned assets meeting at least the minimum needs of the responders. UAS may have provided better info, cheaper (also free in that case) but it’s a hard sell to shut down something that’s working during a disaster to use something new. The FAA was in no way a problem or obstacle, it was a FEMA decision that it wasn’t worth it. And I can understand that decision.

I think most people don’t understand why an area has to be “sanitized” of manned vehicles for a small UAS to be used at altitudes significantly lower than a manned aircraft would fly at. One reason is that the manned aircraft may indeed be flying very low, like a helicopter lower a collar to lift a person out of flood waters. That is extremely dangerous-no-room-for-surprises-or-there’s-a-crash-and-everyone-dies operation.  So if tactical rescue is a possibility, you don’t want anything even nearby that could cause a problem.

I also think that the hobbyist versus public aircraft issue is poorly understood. Due to the quirk of the 1960s when dad (I never saw a mom on these outings) would take the kids, a picnic lunch, and the dog to a field to fly a RC plane, hobbyist got special regulations. Fast forward 50 years and these regulations are letting untrained people fly over crowds with open rotor ‘copters that have not been in any realistic way been tested. Yep, that loophole needs to be closed in an appropriate way. However, if the platform isn’t owned by a hobbyist, it falls under FAA regulations that require a COA. This is not necessarily a bad thing if you’ve been watching the news where ‘copters are injuring people.

But it doesn’t matter who owns what for a disaster because that airspace is totally cordoned off and hobbyist rules don’t apply. You aren’t allowed to just decide to fly your quad rotor or RC plane to help, because you are messing with the response. A manned system has to hightail it out of the area, land, and report what happened if it seems something that isn’t supposed to be in its area. So even if it seems like a good idea, the risk of interrupting a very important life-saving mission that you (or your police friend) knew nothing about is unacceptable. There’s a low risk but a high cost of failure if you interrupt the response. Self-deploying is bad, bad, bad.

Is the current system non-optimal? Definitely! Am I a happy camper? Not particularly. Do I understand why an agency might just say no? Yep. But I urge everyone to be aware of the rules and play by them. I don’t want to be trapped in a flood with a basket being lowered down to me and then the helicopter waves off, leaving me there for hours.

in memoriam: Michael Beebe a Roboticist Without Borders

It is with great sadness that I must report that one of members of Roboticists Without Borders, Michael Beebe, has passed away unexpectedly at his home. Michael epitomized the RWB spirit of our volunteers- he had attended the 2013 Summer Institute at Disaster City® at his own expense in order to be trained as a data manager for the field teams. He has been an active advocate for disaster robotics, leveraging his experience as a Commander in the US Navy and later as a consultant on major initiatives such as Dr. Gary Gilbert’s robotic casualty evacuation program. Mike’s positive attitude, hard work in promoting robotics, and outgoing nature was an inspiration to me. He represents the commitment and expertise that make the Roboticists Without Borders program so special. Please see: for Michael’s Obituary.

UAVs and the Moore Tornado: response to CNN blog

CNN has a nice blog post on the UAVs that weren’t used at Moore.  Here are my comments:

Small UAVs have been used at 11 disasters internationally. The first use of small UAVs was in the US by the Center for Robot-Assisted Search and Rescue, which I direct, during Hurricane Katrina as part of the Florida State Emergency Response Team. We have been advising on the use and procedures for getting permissions for the tornado response, as flying even a small UAV requires coordination with the other activity- hence the no fly zone. The FAA has had an emergency COA process for years, though we find many agencies and industries are not aware of it. We are happy to assist agencies and industries in adopting and deploying unmanned systems of any kind.

OKC Tornado– unmanned systems not the best fit, here’s why

Our hearts go out to the victims and the responders in Oklahoma. We have been working with the FEMA Innovation Team from shortly after the devastation occurred, however aerial and ground unmanned systems are not a good fit for this situation.

In terms of UAVs: There’s already aerial coverage from manned assets and it does not appear that multistory commercial buildings are heavily damaged. Two-story houses and apartment buildings and “strip malls” are well understood failures so additional aerial views are generally not needed to provide more information. If UAVs were available to the first responders, then they would be a much less expensive source of aerial information than manned helicopters or exploiting news helicopters. UAVs provide the ability to serve as wireless nodes (indeed, a big shout out to Roboticists Without Borders member Black Swift Technologies for their work with that) but the coordination with air traffic control and manned assets plus the deployment of COWs (cellular towers on wheels) means that if there weren’t available immediately, they are less likely to be of benefit.
In terms of UGVs: This is a wide affected area with “shallow” debris versus a big building collapse. Canines are the quickest way to locate any survivors that aren’t shouting or aren’t on the surface of the debris. You don’t need the UGV to penetrate the debris further than what a search camera can go to help find survivors or speed extrication.
In terms of UMVs: If there are lakes and streams, marine vehicles might be useful in searching for missing person who may have been swept into a pond and drowned.
We continue to stand by to provide assistance as needed.

Draganflyer credited with first live save with a search and rescue robot!


Check it out here:

Last night (May 9, 2013), a Draganflyer X4-ES UAV  with the FLIR Tau infrared imager was used by the Royal Canadian Mounted Police to locate the driver after a car wreck- he had wandered off in the cold. Dragan is a member of our Roboticists Without Borders and has been active in our CBRNE experimentation. The medical personnel said the driver would have had only an hour or two more to survive.  This is the first reported life saved with a Public Safety UAV. Congratulations!

China earthquake and Bangladesh collapse… the challenges of remote disasters

The Chinese earthquake and the Bangladesh collapse coming on the heels of the Tanzania building collapse illustrate the need for rapidly deployed, regional teams of disaster robots that can quickly get there. The Bangladesh collapse might have been aided by the use of small robots to penetrate in the rubble. Ground robots are less useful for a wide area of residential buildings, though UAVs are very helpful for assessing the extent of damage. But for now, the best we can do in the rescue robot community is to send our thoughts and prayers to the victims, their families, and the responders.

Waco Explosion: no robots needed

We are all keeping the victims, the families, and the responders in the devastating explosion of the fertilizer plant in Waco in our thoughts and prayers. We’ve reached out and there doesn’t appear to be a need for robots at this time. The event occurred in the evening, a problematic time to fly sUAVs to get an immediate overview. Damage  to residential areas do not require robots, as canines are much faster at detecting victims and the debris is usually sufficiently shallow that the interior view from a robot is not needed for searching or extrication. Robots are sometimes useful for forensics- to enter areas and capture the scene before people enter and disturb it by virtue of entering. But the word is that they aren’t needed for this case.

Iran Earthquake: CRASAR monitoring but this type is hard for robots

Dr. Tetsu Kimura and member of our IRS-CRASAR expeditions is coming back from Robocup Rescue in Iran and has asked about CRASAR robots for the Iran earthquake. I am taking the liberty of sharing my reply:

The earthquake is awful and what another tragic loss of life.  I am a great admirer of Amir and his efforts. We’ve been watching the earthquake here– I don’t think the robots would be of much use but certainly would try to support a response. It is painful to see the loss of life.

The primary damage based on the media- which could be wrong- appears to be to mud and brick houses versus multi-story commercial buildings, if victims survive they are probably fairly shallow (less than 6m) and in voids surrounded by brick and mud has become sand– there are generally no voids from the surface to the survivor for the robot to penetrate. Dogs can readily detect the presence of a person and then it requires manpower for extraction. Existing techniques work well for depths of 6m.  Robots are slow compared to canines, and CRASAR deployment with FLTF-3 during Hurricane Charley and FLTF-3 deployment of ground robots at Hurricane Katrina showed that ground robots didn’t provide a cost/benefit for wide area search of urban residences. So unless it’s a multi-story building such as an apartment that has collapsed, current ground robots won’t make a difference and we recommend more canine teams.  In the future, something like Dan Goldman’s sandsnake robots on a large scale could help.

Landslides are also challenging for ground robots, as we saw at the La Conchita mudslides- as with the mud and brick residences you have nothing but dense dirt, not the void spaces seen in a commercial concrete structure.

The nuclear facility is another matter, of course, and the situation may call for ground and aerial inspection.

For the wide area search of residences, besides canine teams other technologies such as ground penetrating radar and better informatics to coordinate researchers and resources would be a huge potential contribution and why the Center for Emergency Informatics exists.

Finally, there is the large travel time as Satoshi noted for the Tanzania collapse, so we would arrive around 72 hours later, outside the probability of long-term survivors. The robots would add little to recovery of the critical infrastructure in this case.

Please let me know what you think. In the meantime our prayers are with the victims, their families, and the responders.

Thoughts on the Second Anniversary of the Tohoku Earthquake and Fukushima


Today is the second anniversary of the 2011 Tohoku Earthquake and Tsunami and the associated Fukushima Daiichi nuclear accident that have taken over 15,000 lives and will cost at least $34 billion.  Our hearts and prayers remain with the Japanese people and all of us at CRASAR are honored to have assisted in a small way.


These terrible events resulted in an unprecedented use of rescue robots provided by nearly a dozen groups to agencies or stakeholders.  A ground robot and a ground-aerial team assisted with earthquake damage inspection. Five marine vehicles were used for tsunami recovery operations. Two unmanned aerial vehicles and at least six different ground robots have been reported at Fukushima. Indeed, Fukushima was notable for the press asking where were the robots? This is quite a change since the first use of rescue robots 10 years earlier at the 9/11 World Trade Center disaster, which saw only six robots from CRASAR used by only two agencies.


Yet, two years later after Tohoku, the use of rescue robots remains ad hoc, with agencies all over the world unaware of capabilities that have been proven in other disasters or in related applications such as inspection or military operations. In the United States, New Jersey Task Force 1 is the only urban search and rescue team with robots; no FEMA urban search and rescue team has a robot of any type.


While agencies largely remain unaware of rescue robotics, the 2011 events appear to have captured the attention of robotics researchers. Based on my observations of the popular press and reviews of articles and grants, disasters have become heavily cited as a potential societal impact for any number of clever concepts in mobility and control–  robot insects and snakes, exoskeletons, UAVs with arms, robots that unfold like proteins, and so on.


Will these innovative lines of basic research in mobility and control really contribute to disaster prevention, response, or recovery? As with all worthy basic research, it’s hard to predict where an innovation will actually be transformative, only that it will be.  The ideas are exciting and I look forward to seeing where they will go in the next 20 years. But I worry: will the hype promoting these futuristic mechanisms confuse agencies as to what is available now? Will the disconnect of these imaginative mechanisms from how disaster preparedness and response actually works and what is needed scare the agencies away?


What is needed is better perception and human-robot interaction. A recent survey of the IEEE Safety, Security, and Rescue Robotics (SSRR) technical committee reported the biggest technical barriers are perception and human-robot interaction, not mobility and control. SSRR, the one scientific community dedicated to disaster robotics, concluded that it is not the lack of mature mechanisms that is keeping robot deployments ad hoc. Rather it is the lack of sensors, image enhancements, interfaces,  visualizations, etc., that is holding back adoption.  Robots often don’t fit into the overall data-to-decision process as they don’t provide enough sensing. They are often designed to simplify the workload on the operator, ignoring the needs of the multiple decision makers who need to see the data from the robot in order to extract information specific to structural integrity, environmental quality, survivor health, etc. The agencies can’t buy systems they can drive but can’t effectively use.


The legacy of the first 10 years of rescue robotics is that the Public expects to see robots at disasters. I hope the legacy of the next 10 years is that agencies will have the robots on hand and the ability to use them effectively.


The CRASARinfo channel at YouTube has video from some of our efforts at the Tohoku tsunami, see

Reuters video: Fukushima disaster tests mettle of local robot makers

Check out this 3 minute video on Japanese robots being used, or developed, for Fukushima. Big shout out to Prof. Eiji Koyanagi at the Chiba Institute of Technology- he’s been a real pioneer in rescue robotics.