Year: 2012

Anniversary of 9/11 and human-robot interaction

It’s 9/11 and it is perhaps karma that I am at the NDIA Human-Systems Division Workshop- because the lesson learned at the WTC was that the robots were physically good enough, but the biggest area for improvement was the human-robot interaction. Check out Jenn Casper’s paper
at http://tinyurl.com/9dt4nrg – it’s one of the most highly cited papers in human-robot interaction.

Our thoughts and prayers remain with the families of the victims and all the people working in homeland security and homeland defense to prevent terrorism and eliminate the need for rescue robots for that particular application.

Iran Earthquake

The death toll for the latest Iranian earthquake appears to be settling in around 300 lives lost- a sad number. Iranian rescue roboticist and colleague Amir Soltanzadeh posted on facebook that most of the victims were in 1 story houses which were easy to search. This is where dogs excel- they can cover large areas, determining if there is a survivor in a house in minutes (or less), whereas a ground robot has no comparable sensors and would have to physically worm through the rubble to visually find a survivor- a process that takes time and energy (as in battery power is limited).  1 story houses are also easier to work with for extrication- a ground robot doesn’t offer much benefit.

Unmanned aerial vehicles may be the best initial choice of robots for disasters that pose extreme scales of distance and area– their first look can help professions prioritize the response and estimate the resources that will be needed. Together with social networking from those on the scene, professionals can get better information must faster.

The Iranian people are in our hearts and prayers and we are all grateful this earthquake was no where as bad as Bam.

Landslide in British Columbia: how robots can help in such events

The sad news of the mudslide in Canada is very similar to the 2005 La Conchita mudslide, described in this paper on rescue robots for mudslides, where CRASAR had its first post-World Trade Center deployment of rescue robots at the request of the Ventura County Fire Department. Mudslides are fluidized, so like water, the mud penetrates everything nook and crevice. Survivors are generally found in the collaterally damaged structures on the periphery rather than in the direct path. Small ground robots can be useful for trying to get into the crushed and twisted houses and buildings, either from the roof or from under the foundation. But robots and unattended ground sensros can also be useful for monitoring the mudslide- because the responders have to worry about the slide breaking loose and sliding more. Everyone had to evacuate La Conchita because of that. Work has been done by various groups to create unattended ground sensors that can be stuck in the ground  of sensitive areas and wirelessly report soil water content (hey– things are fluidizing here!) and movement (hey- I’m beginning to creep and shift, big movement may follow).  One idea is to use aerial robots to drop these networks of sensors in place after a disaster to help monitor. Otherwise, geologists have to periodically laboriously climb up (and hope not to trigger more slides) and take manual measurements. Our prayers go out to the families and the responders.

Below are pictures from La Conchita:

637167

On the use of construction robots at the Elliot Lake Mall Collapse

Nice article here on the use of mining and construction robots at the mall collapse in Canada. However, the article’s promotion of heavy-duty machinery for search, versus for extrication, may be misplaced.

Our work at 15 disasters since 9/11 and documenting the other known responses strongly indicates that for the search phase, very small agile robots with 2-way audio are desirable. They are small enough to get into the irregular voids or be lowered in through the roof, they are light enough not to cause a secondary collapse, the can move around and get better viewpoints than with a search cam,  inexpensive, and easily transported (from the back of a truck into a backpack…).

A recent example of this is the Hackensack New Jersey Prospect Towers collapse where the NJ Task Force 1 and the UASI teams used Inuktun robots to search for survivors with a couple of hours of the incident. Inuktuns have about a 300 ft long tether, a search cam is usually only to penetrate 18 feet.

Big, heavy gear is certainly of great value for removing rubble, bracing structures, etc. It’s just not the same as small robots for search, finding and interacting with the victim until they are extracted (which can 4-10 hours).

News video: Our aerial and ground robots with Austin Police Department

We’ve had members of the Police Technology Unit of the Austin Police Department over to Disaster City twice to see our Dragan and AirRobot UAVs and our Packbot 510 and SUGV.  SPO Eric Cortez and his colleagues have been terrific in helping us learn more about how they would use these devices, what they look for in a disaster (which the new Dr. Josh Peschel worked on for his PhD),  and how fire rescue and police might share and coordinate. Here’s the link to the Austin Fox Channel video– one embarrassing aspect: Disaster City is a Texas Engineering Extension Service (TEEX)  facility not the Center for Robot-Assisted Search and Rescue.

75th Anniversary of the Hindenburg

The 75th anniversary of the Hindenburg disaster was yesterday, which leads me to a shout-out for New Jersey Task Force 1 which is housed at Lakehurst Naval Air Station in old blimp hangers within an easy stroll of the site of the fire (there’s a nice plaque marking the spot). New Jersey Task Force 1 was the first US team to adopt rescue robots, almost immediately after 9/11. They continue to explore new technologies such as new sensors and small UAVs. Keep up the good work!

Thoughts about the DARPA Grand Challenge…

I’m getting bombarded with emails about the incipient DARPA grand challenge in disaster robots– very exciting- both the idea and the attention rescue robotics is getting!

While I haven’t gone through the BAA in detail (the whole email barrage thing plus I teach on Tuesdays), the media coverage and speculation highlights 3 things that I especially like:

The idea of integration is fantastic and a key enabler in making robots adoptable. Since 1999, we’ve seen this gap between an interesting sensor or mobility platform and the “full meal deal” of working in a scenario.

Another interesting idea is the use of humanoids. Up until Fukushima, rescue robots have been primarily used for sub-human scale space– spaces where people simply couldn’t go because they didn’t fit. Fukushima and indeed chemical disasters such as Bhopal occur in human-scaled spaces, where people can physically fit but may not be able to survive or work long (or well) with protective gear. The rule of thumb is that robots don’t replace people or dogs, they do things that humans can’t do or can’t do for long enough or well enough– hence our name: Center for Robot-Assisted Search and Rescue. Through funding by the National Science Foundation, we’ve been working with TEEX on human-robot interaction for chemical, biological, radiological and nuclear events- and we see huge possibilities for land, sea, and aerial robots.

And it’s win-win: the impact of improved manipulation would benefit robots operating in either scale of space– the skills that allow a a large robot to open doors could be used by a small robot to move rubble out of the way or help triage an unconscious victim.

The focus on the media appears to be on humanoids, which I hope doesn’t detract from other types of mobility or modalities. There are often aerial and water-based aspects of disasters- at Fukushima, Westinghouse used the Honeywell UAV to sample radiation and get close up views of structural damage (I assisted the Westinghouse team). Marine robots could have been used to monitor pollution in the sea.

And keep in mind that from a robotics perspective, there are at least 12 very distinct activities for rescue robots beyond the direct intervention needed to have prevented the explosions as Fukushima. These are search, reconnaissance and mapping, rubble removal, debris estimation, structural inspection, in situ medical assessment and intervention, medically-sensitive extrication and evacuation of casualties, acting as mobile beacon or repeater, adaptive shoring, logistics support, victim recovery, and serving as a surrogate for a team member. This list was compiled based on feedback from responders and what they’ve asked for or speculated on based on our 15 deployments and 30+ exercises we’ve participated in.

A good starting place is Chapter 50 Search and Rescue Robots in Handbook of Robotics and I’m working hard on my forthcoming book on Rescue Robots for MIT Press.