|Fig. 1: The PackBot from iRobot. (Source: Wikimedia Commons)|
We live in a day and age in which robots build our automobiles, delicately perform surgery deep within the body, and repair oil leaks nearly a mile below the sea surface. One would expect robots to play a linchpin role in the response to the Fukushima Nuclear Power Plant accident of March 2011. Yet just months on the scene, Quince, Japan's $6 million robot, became trapped in the confined spaces of the power plant.  Two years after the accident, it is still a hostage to Fukushima.
Notably, at least three robots have been sent to Fukushima from iRobot, a U.S. company most famous for its household robotic vacuum cleaners. We have a special glimpse into the front line rescue operations of these robots through the blog of "S.H.," a Fukushima robot operator who candidly documents his daily work. Although once publicly available, the blog has since been shut down for reasons not made public. Fortunately, an IEEE Spectrum writer has archived the website and published excerpts of it on the publication's website. 
Why is repair work in a nuclear plant a few hundred meters away so much more challenging than stopping an oil leak at the sea floor, a thousand meters from the surface? For one, the plant is strewn with rubble, making access difficult even for rescuers. Although robots can be sent into these dangerous areas without risking human lives, a stuck robot like Quince is more than just costly - it can impair access for other robots in tight spaces like stairwells or doorways. It may be tempting to relegate details like stairwell access to the realm of practicalities that can be overcome by a skilled operator, but in fact, many of these details can easily grow to become major sources of delay or complete showstoppers. S.H. writes with pride of how he was the first to successfully open a round door knob with a robot. When iRobot sent the Warrior, a larger robot, S.H. pointed out the difficulty of navigating narrow stairwells and humorously photographs holes the robot punched in the drywall as he was practicing. 
Beyond the difficulties of dexterous locomotion and manipulation, nuclear response robots face the twin challenge of intense radiation exposure and unreliable wireless communications. Radiation can damage electronics in two ways: by physical damage to the semiconductor crystal structure by near field neutrons and by shifting electrical charges due to ionizing gamma and X-ray radiation.  In either case, the operating characteristics of individual electronic components shift, resulting in device failure. So-called radiation-hardened devices are tested by measuring the total dosage (often in Sieverts) they can withstand before malfunction. But since radioactive damage is statistical, device survival is never guaranteed. S.H. notes in his blog that the video image from the robots became distorted as he piloted the robot near radioactive hot spots. 
Of course, reactor electronics must be designed to be radiation-resistant as well. This is accomplished by a combination of radiation-hardening and shielding. The latter approach causes communication challenges for rescuers as the heavy concrete and lead structures of a reactor limit wireless reliability. S.H. describes a repeater strategy the team improvised to extend the communication range of a robot: a helper robot is stationed part way in the plant, tethered to a fiber optic cable so that the first robot has extended radio range deep within the plant.
As one would predict, the challenges are not strictly technical either. S.H. documents frustrations with the sometimes erratic planning by his superiors and personality conflicts between the front line workers and the distanced management. He even recounts a day where an employee from a different branch of Tokyo Electric Power drives a truck over a robot's communication cable, risking a severed cable and a second stranded robot. Traffic cones clearly marked the cable and S.H. told the truck driver he could not drive over it, but he was deliberately ignored. Such incidents are an inevitable part of complex disaster response highlighting the importance of effective management and respect between all groups involved.
All of these struggles raise a larger question over the design of nuclear reactors. Artificial intelligence pioneer Marvin Minsky writes, "I am appalled by the nuclear industry's inability to deal with the unexpected...The big problem today is that nuclear plants are not designed for telepresence."  In contrast to a nuclear power plant, many of the valves and actuators on deep sea equipment have been specially designed for easy use by robotic manipulators. Automobile manufacturing plants are now designed with robotic integration as a top priority and medical devices are now specially designed for robotic platforms. Little has changed in the robot-readiness of nuclear power plants since Three Mile Island thirty years ago, when three robots were designed by Carnegie Mellon University for survey and decontamination of the unit 2 reactor.  Interestingly, the third robot, dubbed Workhorse, was intended to help decontaminate and disassemble highly radioactive structures inside the plant, but it was never deployed due to cost and complexity concerns.
What explains the lack of progress? No certain answers exist, for this is a topic deeply intertwined with public perception and politics. Eiji Koyanagi, vice director of the Future Robotics Technology Center in Japan hypothesizes that funding for Japanese nuclear response robotics dried up after the 1999 Tokaimura accident, because the country was trying to protect the impression it had painstakingly created of the near-absolute safety of nuclear power. Koyanagi says funding such research would have meant that, "people were obviously going to ask, 'Wait, is there going to be a situation so dangerous that humans can't enter the plant?'".  It remains to be seen if this attitude will shift after Fukushima as Japan struggles to rebuild its confidence in a vital source of its energy.
© Kevin Mori. The author grants permission to copy, distribute and display this work in unaltered form, with attribution to the author, for noncommercial purposes only. All other rights, including commercial rights, are reserved to the author.
 M. Saito, K. Takenaka and J. Topham, "Insight: Japan's "Long War" to shut down Fukushima," Reuters, 8 Mar 013.
 E. Guizzo, "Fukushima Robot Operator Writes Tell-All Blog," IEEE Spectrum, 23 Aug 2011.
 R. Courtland, "Radiation Hardening 101: How To Protect Nuclear Reactor Electronics," IEEE Spectrum, 22 Mar 2011.
 M. Minsky, "Telepresence: A Manifesto," IEEE Spectrum, Sep 2010.
 D. Lovering, "Radioactive Robot: The Machines That Cleaned Up Three Mile Island ," Scientific American, 27 Mar 2009.