CRASAR back from Japan: 6 sites, 5 days, 4 new research areas, 3 robots

CRASAR-IRS team photo, courtesy of CRASAR and NSF (Robin Murphy, Karen Dreger, Eric Steimle, Sean Newsome, Tetsuya Kimura, Jesse Rodocker, Satoshi Tadokoro, Kenichi Makabe)

We’re now all back in the US. Our third day at Rikuzen Takada, in the pouring rain and high wave activity, did not find any victims, so we were a bit disappointed. We participated in a press conference held by our hosts, the International Rescue Systems institute at the Chiba Institute of Technology on Sunday morning (Dr. Anne Emig from NSF Tokyo was there and it was great to meet her in person- she’s been a tremendous help), then dropped off gear at Continental cargo (huge thanks!), boarded our flights, and flew home. We cannot thank our host Prof. Tetsuya Kimura, Prof. Fumitoshi Matsuno, their grad students, and the IRS team enough for their help! We hope to obtain NSF funding to return with a different set of robots better tuned to searching for bodies under flotsam or underwater debris within a month.

It was nice to see that Minami Sanriku Mayor Hitoshi Sato publicly stated that the new port facility was being reopened because of our search.  The Rikuzen Takada City Manager, like the Minami Sanriku fisheries expert, also was interested in the ROVs for use with fishing and oyster farming- a positive, unintended consequence of being there.

Here’s some numbers about our work:

In total we searched six sites in Japan over 5 days.

Of the four robots in our cache, we only used 3. We used the SeaBotix SARbot at each site- it was definitely engineered for underwater search and rescue making it easy to quickly deploy. The SeaBotix LBV-300 was not used, as the SARbot was sufficient for the areas of interest and using the LBV-300 at the same time as the SARbot would require them to have to work in different areas to avoid their sonars from interfering with each other.  The Seamor was used twice and its DIDSON sonar (which doesn’t interfere with the SARbot Gemini sonar) had some advantages but wasn’t intended to be a rapidly deployable system. The Access AC-ROV, essentially a camcorder with thrusters, was also used twice in clear water to assess debris in very shallow water. It was fun to literally throw it in the water.

At the press conference we reported out on four preliminary findings on needed research:

1. Simulation and Geographical Information Systems- we’d like to see work in projecting the location of victims, and mechanisms to then update models as bodies were recovered

2. Cyber-physical systems- we definitely see the need for autonomous station-keeping to keep objects in view and accurate control in all kinds of water conditions
3. Computer vision & cognitive engineering- while underwater search and recovery is something that is perceptually a bit like Supreme Court Justics Potter Stewart’s quote about pornography (“I’ll know it when I see it”), we believe that cooperative computer vision algorithms permitting object cueing (“look here dummy”)  and  sensor fusion of sonar and video would be of great benefit.
4. Human-robot interaction- I suspect based on my rough observations that the operators never got beyond Level 2 Situation Awareness using Dr. Mica Endsley’s scheme, so help with visualization of surrounding clutter and advances in multi-modal interfaces would be useful.

Comments are closed.