All posts by Dr. Robin Murphy

About the author:

avatar I am a founder of the field of rescue robotics; my personal goal is to enable emergency responders to routinely save lives through physically situated artificial intelligence (robots, sensor networks, etc.)

9/11: Thoughts on the Anniversary of the WTC Disaster and First Use of Robots

I am spending the anniversary of 9/11 at the World Bank Headquarters at the World Reconstruction Conference. I was invited to give a talk on how disaster robots can be used for the recovery phase of disasters (as opposed to the search and rescue/response phase). In many ways, it was the kind of talk I had expected to give on the 13th anniversary of the first use of robots for a disaster.  I was able to proudly show  that land, sea, and aerial robots are already being used for recovery efforts. For example, the joint IRS-CRASAR team that fielded marine robots to the 2011 Japanese tsunami helped with the recovery of the region, re-opening the Minami-sanriku fishing port and finding polluting debris in the aquaculture. The decommissioning of the Fukushima Daiichi nuclear power plant is likely to go on for decades and robots are essential to the safe and cost-effective work.

But it was also the kind of talk that I hadn’t expected to give because 13 years after the successful use of ground robots, 10 years after marine vehicles, and 9 years after small aerial vehicles, robots still are not routinized into disasters! The responders don’t have them and as best I can tell in 36 disasters where robots have been reported to be used, the robots were borrowed in 35 cases- the agency that needed them didn’t have them.

The past 13 years have continued to show the potential, I believe the next 3 will be where we see the rapid adoption of disaster robotics.

Our respects to the victims, their families, and the responders and my thanks to the great team that John Blitch pulled together for CRASAR’s and the world’s first use of robots.

China earthquake: Drone on-site plus other ways robots can help.

YouTube Preview Image The death toll continues to rise in China’s Yunnan Province earthquake and our thoughts and prayers are with the victims, families, and responders.  The Chinese Army is already using drones to provide responders (as well as geologists and hydrologists) with assessment of damage in the remote, rural areas. In general, ground robots are not terribly useful in earthquakes unless there are significant building collapses where canines can smell survivors but robots are needed to crawl in spaces that responders and canines can’t get in.  Unmanned marine vehicles can be useful in help inspecting the underwater portion of bridges and rapidly determining river channels are open so that ships can bring in supplies.  So while robots are unlikely to directly save lives, they fulfill their goal of helping the responders help the survivors!

Robot used to search from Queensland miner

story here

Colorado Mudslides: UAVs and Roboticists Without Borders on standby

The Colorado mudslides appear to be the Washington state SR530 mudslide writ large (4 miles long versus 1 mile long), though thankfully with a search for three people, who could still be alive versus the 43 killed in Washington state.

rwob patch.pptxMesa County is the best place in the world to have a mudslide- Ben Miller, in the Mesa County Sheriff’s office and now director of its Unmanned Aircraft Program, has been an early adopter of small unmanned aerial systems (sUAS). Under his direction, Mesa County got the first approval for an agency to fly over an entire county. His “flock” includes a Draganfly X4-ES rotorcraft (Draganfly is a Robotocists Without Borders member!) with their own version of advanced 2D/3D mosaic software that the geologists and hydrologists at SR530 found so useful. Ben also has a Gatewing and Falcon fixed-wing sUAS.

The UAS providers of the Roboticists Without Borders team (Black Swift, Draganfly, Precision Hawk, and Texas A&M) are on stand-by to assist, possibly providing a LIDAR platform and additional software.  Precision Hawk and their geospatial software, of course, were the stars of our SR530 mudslide response providing an interactive 3D reconstruction of the “moonscape” in less than 3 hours of processing time on a laptop.

Black Swift has been doing some phenomenal work that could prevent mudslides and flooding- they are developing a miniature microwave radiometer package for NASA for their sUAS that can detect soil moisture– which can determine if the soil is saturated and thus about to flood or slide. The package isn’t ready yet, but think about the implications for being proactive next spring!

The size of the mudslide raises the question of the use of multiple sUAS in a divide-and-conquer strategy. There has been a significant amount of research on this in terms of optimal path planning and general coordination. I believe the University of Colorado Boulder may hold COAs by the FAA which permit multiple platforms to be in the same area at the same time– for their storm formation studies, but I could be wrong.

Please donate to Roboticists Without Borders so that team members can continue to donate their time and equipment to help responders and accelerate the adoption of the technology.

 

Preventing disasters: small unmanned aerial vehicles for evacuation and crowd control

This month’s issue of Smithsonian Magazine has an article  “Why are people so comfortable with drones?” on  Brittany Duncan’s preliminary study for her NSF graduate fellowship research on using small UAS for evacuation and crowd. Brittany is my PhD student who recently flew the AirRobot at the SR530 mudslide response. It’s nice to see that robots are being considered for more than the immediate life-saving aspects of search and rescue. Brittany sees a near future where aerial vehicles can act as “headers” and “heelers” to guide and block people into following the right exits during an evacuation.

Soma Turkey Mine Disasters: Mine Disasters and Robots

(note: this blog was referenced by New Scientist) The horrendous number of victims at the Soma mine disaster in Turkey continues to grow. We immediately reached out to Turkish officials through Dr. Hasari Celebi at the Gebze Institute of Technology as a guiding force.  Less than a year ago I had given a keynote at a workshop on disaster robotics being held there by the government with Dr. Celebi as a key driving force.  It was clear that there was many scientists interested in applying their great ideas to earthquake response. Unfortunately it takes time and sustained investment to create a robotics capability for handling such a tragedy and the mine disaster was too soon.

Most people don’t realize that mine disasters have been to date the most common situation for ground rescue robots. As I note in Disaster Robotics out of the 28 disasters where robots have been used or on site between 2001 and 2013, 12 (42%) were underground mine incidents. Of the four disasters where robots were on-site but could not be used, 3 (75%) were underground mine disasters- showing just how much this technology is needed.
2 MSHA wolverine Underground mine disasters in coal mines are especially challenging as there may be methane to cause explosions or because the mine itself catches fire which is difficult to suppress. Mine responses are challenging based on the type of entry into the mine. Dr. Jeff Kravitz at the Mine Safety and Health Administration (MSHA) is the expert on robots for underground mine disasters and we co-authored an article “Mobile robots in mine rescue and recovery” in IEEE Robotics and Automation Society Magazine that summarizes the opportunities and challenges for robots based on an analysis of deployments in the US. MSHA may have the only mine permissible robot in the world (i.e., certified not to cause an explosion in a methane-rich environment)- the V-2, an Andros Wolverine shown in the photo.

 

CRASAR has assisted with two mine disasters Midas Gold Mine (2007) and Crandall Canyon (2007) and was requested to assist with two others but the technology wasn’t there. The Midas and Crandall Canyon events used smaller robots, an Inkutun Xtreme loaned from the pool at SPAWAR and a Inuktun Mine Crawler that could go down narrow boreholes. We also conducted a report for MSHA on underground rescue technologies that allowed access to their use of robots at 7 other mine disasters.

Our thoughts and prayers are with the Turkish miners, their families, and the responders.

(Updated with video) Flew UAS at SR-530 Mudslide

AirRobot flying moonscape at SR-530 mudslide April 23, 2014

AirRobot flying moonscape at SR-530 mudslide April 23, 2014

Roboticists Without Borders returned with member FIT to Washington state with platforms from CRASAR and PrecisionHawk members to order to help determine the eminent risk of loss of life to responders, as they continued to work downslope of a potential secondary mudslide or a breach in the river. Many people assume that disaster robots are just for immediate search and rescue of survivors, but this is one of many examples of where robots can protect the responders.

Our missions were collecting data for the geologists and hydrologists from the “moonscape” and toe of the river where it was impossible to manually survey due to the flooding and quicksand-like mud and couldn’t be surveyed from manned helicopters or see from remote satellite sensing due to the higher altitudes and less favorable viewing angles. These areas are next to the cliff face of the mudslide and not in the victim recovery area.

We flew the AirRobot 100B platform under an emergency COA from the FAA on April 23  but the high winds in the narrow canyon prevented us from flying on the 24th. The PrecisionHawk was not granted an emergency COA, but we used the PrecisionHawk software to do 2D tiling of imagery and to create interactive 3D reconstructions which I will post soon (it’s finals here at Texas A&M). Brittany Duncan and I collected about 33GB of data in 48 minutes of flight time covering 30-40 acres with the CRASAR AirRobot and then about 3 hours of post processing on a laptop by the PrecisionHawk team (Tyler Collins and Justin Kendrick). Getting this type of data for ESF#3 and ESF#9 functions often takes days– now it can be done by them on demand.  This is revolutionary!

FIT has a press release here and I’ll be posting photos and snippets. Big shout out to FIT who helped support the mission with both personnel on-site (Frank Sanborn and Tamara Palmer) and with partial funding.

Speaking of funding– our deployment war chest is empty. CRASAR pays for travel, PPE,  etc. whenever possible for our volunteers, breakage and software upgrades, and this drained the last of our funds. We’re setting up online donations so that you  can join RWB as a funding provider and donate to the cause!

YouTube Preview Image YouTube Preview Image

Oso Washington Mudslides– We had the UAVs there but didn’t fly

Just returned from the Oso, WA, mudslides with the Field Innovation Team (FIT) but didn’t fly due to drone privacy concerns from Snohomish county. The upside is that we now have a template for manned/unmanned airspace deconfliction and can assist others in getting emergency certificates of authorization (COAs ).

We had requests from the county to fly small UAVs first thing on Thursday but it was Friday morning before we had three assets on site: two fixed wings (the Insitu ScanEagle, Precision Hawk Lancaster) and one rotorcraft (our Air Robot 100B). All of these were provided through our Roboticists Without Borders program at no cost to the county, with Insitu and Precision Hawk diverting their teams led by Kevin Cole and Pat Lohman respectively from their current jobs.

The reason for UAVs was straightforward. Responders such as WA-TF1 and WA-TF4 working on the rescue and recovery are at great risk from even a small slide or flash flooding as the river is continuously changing and ponding as the rains continue. The site itself is gooey mud and workers would have to be evacuated by helicopters hoisting them out. The canyon is narrow with trees and thus it is hard to get complete imagery from manned assets to predict landslides or manage the flooding. Geologists are gently swarming the edges of the slides setting up sensors but there is still some visual information missing. The ScanEagle and smaller Precision Hawk are world class for geospatial reconstruction and flooding. FIT had arranged for post processing of the AirRobot quadrotor imagery with new 3D reconstruction software from Autodesk. Chief Steve Mason, West Division, talks about the potential for UAVs in this article. We also got a shout out in general.

We worked with the Engineering Branch to determine flight paths and payloads to monitor the river flooding and to get 3cm per pixel higher resolution scans of the lower slide, the cliff face where geologists where having to rope themselves off to take measurements, and the “moonscape” area of the slide currently inaccessible by foot and thus the response teams couldn’t plan how to access. The team went with the Engineering Branch out to the site to refine the missions, identify launch and recovery areas, and how to maintain constant line of sight.  Insitu had to pull out on the first day because we couldn’t find a satisfactory launch and recovery space within the Temporary Flight Restriction (TFR) area, which is airspace over the incident.

The need for the manned helicopters to fly at a moments notice for emergency evacuation combined with regular manned missions and the narrow canyon presented some challenges. Manned helicopters are extremely vulnerable when flying at low altitudes, and even a large bird can take them down, see the article about a near miss. That’s why a TFR is set up and it automatically bans any aircraft including UAVs operating under hobbyist rules. Everything has to be coordinated through the Air Branch of the incident.

The Air Branch instead of just saying “no” to UAVs did the opposite—they welcomed us and did a fantastic job of coordination, with Bill Quistorf helping us create a jargon-free airspace deconfliction plan that should work for just about any incident. Randy Willis and Mark Jordan at the FAA stayed on call through out the weekend working on the emergency COAs for each platform.  Chief Harper gave us a room in the Oso Community Center next to the Oso Fire Station to stage in and we were touched by the generosity and community spirit of everyone we met.

However, as the Operations Branch put in the formal request for the finalized missions and we got ready to fly late on Saturday, Snohomish County Emergency Management in Everett stepped in and blocked the request. After discussions on Sunday, the new Incident Commander Larry Nickey cancelled the missions based on concerns about privacy. The families were already worried about the media leaking photos and some were very contentious about drones. It was at that point that I recalled that Washington state has some of the most restrictive drone anti-privacy laws in the country, so there is already distrust in general.  There was no way for the Engineering Branch to determine without flying the UAVs if the data would be sufficiently better than what they were getting now and would significantly increase the safety of the responders to justify overruling the families. This just wasn’t the time to go into the chain of custody of the imagery or that these were no different than imagery than from the unmanned systems; the families in their grief can’t hear and the EOC personnel shouldn’t be distracted continuing to push for activities that make the families uncomfortable. Larry made the tough, but understandable, call to cancel the missions, but left the door open for flights after victim recovery was complete and the activity was cleanup and reopening the area.

This was the first outing of Roboticists Without Borders with FIT. FIT is lead by Desi Matel-Anderson, former Chief Innovation Advisor for FEMA, Rich Serino, former FEMA Deputy Administrator, and Tamara Palmer, former Program Specialist with FEMA’s Recovery Directorate, to help communities get innovations that they may not be aware of or know how to access. FIT staff worked to understand the needs of local officials and connect us with Autodesk.  Frank Sanborn served as the FIT coordinator for us and he and Stacy Noland get big shout outs for helping with everything from lugging gear and driving 120 miles on Sunday between Everett, the EOC, and the site to try to get the mission request unstuck.

We are disappointed that we can’t help out but our hearts and prayers go out to the families and all of the fine people working this very, very tough event.

Washington State Mudslides Highlight Challenges for Ground and Aerial Rescue Robots

All of our thoughts and prayers go out to the victims, families, and responders. 

I am already being asked about how robots could be used for the horrific mudslides in Washington State. 

To the best of my knowledge, robots have been used only once for a mudslide. That was the CRASAR deployment in 2005 for the La Conchita, California, mudslides which provided unmanned ground vehicles at the request of Ventura County Fire Rescue and Los Angeles County Fire Rescue. Direct victims of mudslides and avalanches rarely have survivors because the ground and snow acts like a fluid displaying the oxygen, leading to suffocation. Victims of collateral damage have a better chance of survivor. CRASAR was called in to help search the crushed houses for missing neighbors; as detailed in Disaster Robotics, the small ground robots didn’t get far in the gooey mud. 

A major challenge for a slide or avalanche is that the robot needs to burrow through a “granular space.” Instead of going through a hole where the hole is at least as big as the robot, the robot needs to go through a space where the holes are smaller than the robot. There’s some interesting research that Dan Goldberg at Georgia Tech, Howie Choset at CMU and I have proposed on exploiting Dan’s work with robots that mimic sandfish, Howie’s miniature snakes for granular spaces but we’ve yet to hit on funding. 

Small UAVs could play a beneficial role. Certainly having the first responder on the scene driving down the road to the slide could able to get a quick overview of how far it extended. At La Conchita we had seen the possibility of small UAS dropping sensors in the slide that could be remotely monitored rather than sending geologists to periodically climb up the slide to make measurements. 

However, the pictures in the news of manned helicopters to airlift out survivors illustrates why air space coordination is a disaster remains a must and why civilians robot enthusiasts shouldn’t fly without permission, like the person filming the Harlem building collapse. The manned helicopters are working in the same under 400ft elevation that many small UAS companies advertise that their systems work in. It is important to remember that if a civilian flies in the airspace over a disaster, their AMA insurance is void (assuming they have insurance) and  regardless they may be subject to legal action. Worse yet, a manned helicopter conducting an airlift has to abort the mission if an unaccounted for vehicle enters their area of operation and thus could cost a victim their life. 

East Harlem Building Collapse: role of ground robots

It doesn’t appear that small ground rescue robots are being used to assist in the search and rescue of the terrible East Harlem building collapses-if you know of any being used please let me know. We offered our robots to DHS the same day and also New Jersey Task Force 1 and the New Jersey UASI team have small robots that they used in the 2010 Prospect Towers parking garage collapse. All of our thoughts and prayers go out to the families of the 7 victims and missing- as well as to the responders working their way through this disaster.

Ground robots have been used  8 times between 2001 and 2013 for search and rescue in structural collapses- crawling underground in building collapses and mine disasters since the 9/11 World Trade Center collapse.  For building collapses shoeboxed sized robots such as the Inuktun VGTV are a popular choice because that are small and the power/communications tether serves as a belay line for lowering the robot. Ground robot have video cameras but can often be outfitted with a thermal camera. The thermal camera is useful for looking for heat signatures of possible survivors and also smoldering fires or live electrical outlets. A robot typically needs both because thermal radiation produces a fuzzy image, not always good for navigation or for structural assessment. We’ve often velcro-ed a thermal camera to a robot and run a separate tether.

Gas leaks are similar to mine explosions in that there is a worry as to whether the electronics of the robots will set off another explosion. The is referred to as whether the robot is “intrinsically safe.” There are different standards for intrinsically safe depending on the industry so that makes it harder for a robotics company to create a certified explosion proof robot.  I know of only one robot that is certified as intrinsically safe- the Mine Safety and Health Administrations very large “sumo” V2 robot, a variant of the Remotec Wolverine.