Wednesday, March 19, 2008

'In the presence of mine enemies...'
-On wild mongoose chases-

An expatriate student is at once an intrepid explorer and an unwitting exile. Having come to a distant land, one must first of all 'make oneself at home' even before trying to 'make a mark'.Be they visiting scholars in exchange programs, professionals on a sabbatical, recipients of scholarships or simply students enrolled for higher education programmes, they are not only faced with the challenges of meeting their personal quests and beating their personal bests; but are often fazed by the keen awareness that their efforts and earnings are owed to and anxiously expected by sponsors, employers, parent institutes or perhaps parents. For someone like me who quit the clamour of a conventional workplace in my home country, and now on a different quest in the conducive calm of university environs, these concerns are best expressed in the words of Kahlil Gibran, "A seeker of silences am I, and what treasure have I found in silences that I may dispense with confidence?". Having 'come all the way' to a distant land, one is certainly not expected to return empty-handed. I have found it heartening to read about expatriate university alumni of US universities who return to their home countries to dispense their newly discovered treasures with confidence, often by single-handedly pioneering a certain field of research in their countries and sometimes founding research institutes. Today I found an outstandingly inspirational example of what it means to share treasures with one's countrymen.

Dr. Thrishantha Nanayakkara* was a guest speaker today at a talk entitled, "Biomimetic Legged Locomotion and Odor-Guided Behavior for Humanitarian Landmine Detection". I will not expand too much on the title, for my motivation for writing this is not as much the talk but the man himself and what drives him. It will suffice to state that a demining squad needs 'biomimetic legged locomotion' because animal legs are much better than automobile wheels while negotiating densely forested territory where mines are commonly found,that they are interested in 'odor guided behavior', motivated by sniffer dogs who continue to be the most tried and tested means in explosive detection, that 'humanitarian landmine detection' does not mean clearing a minefield by blowing it up and leaving a hole in the earth, but reclaiming the land unscathed. Coming to the man; Dr. Nanayakkara, currently a visiting researcher at Harvard, hails from Galle in terrorism-ravaged Sri Lanka and has a multidisciplinary academic background including a bachelors in Electrical Engineering from University of Moratuwa in Colombo, a doctorate in Systems Control and Robotics from Saga University in Japan and post-doctoral research in Biomedical Engineering at Johns Hopkins University. As a background for today's talk he gave a brief bloodcurdling account of ethnic violence in northern Sri Lanka raging over the past three decades. Landmines, a staple for guerrillas the world over, are 'indiscriminate weapons' built not to target a known enemy but to wreak blind destruction on perhaps a hapless toddler. Children joyfully treading to village schools have fallen prey to landmines, scaring their classmates from taking the same route the next day, robbing them of childhood education and thus making them susceptible to the canvassing of militant groups. Breadwinners on their way to the workplace have met their end in landmine explosions, leaving families destitute and again susceptible to terrorist recruitment. In four simple words capturing his patriotic and humanitarian concerns, Dr Nanayakkara summed up his description of landmines by simply saying with a grim, sardonic sigh, "I don't like them!". But he did not stop by just 'not liking them'. He chose to draw upon his education in Electrical Engineering, Robotics and Biomedical Engineering not merely to climb pedestals of publications and patents in a pedagogic world; but move right to the war-zone as it were in his professed mission: to rid his homeland of landmines within 15 years. Enriched by an international education, the treasures he wishes to dispense among his countrymen are not treasures shipped from a foreign land. Rather, he is reclaiming for his people the bountiful earth that is their own, but which they aren't able to farm; the schools and shrines which are theirs but languishing in ghost towns. What he is doing is not just dispensing treasures with confidence. From the fiery landmine-infested hell that his homeland has become, he is doing nothing less than reclaiming, fearlessly and resourcefully, the Paradise on Earth which he and his countrymen are justly heirs to. May his tribe increase!

Besides the obvious humanitarian and patriotic passion which Dr. Nanayanakkara brings to his work, the research** itself is cutting-edge and literally operates at the frontiers the robotics and computer science fields(while its results are operational at frontier mine-fields!) Just to give a hint of why this work ought to excite engineering researchers at large, I will just paste here what USC's Viterbi School of Engineering's online announcement listed as keywords for the talk, "Keywords: Field robots for landmine detection, animal-robot cooperation, adaptive control, reinforcement based learning, fuzzy and neural network based control, evolutionary optimization." The requirements of any solution to the de-mining problem is that it must necessarily be unmanned, capable of mobility on soft muddy terrain, capable of navigation through dense vegetation and be able to locate a target with a signal gradient analogous to the sniffing of a dog. I will dwell briefly upon a thought that struck me when Dr. Nanayakkara mentioned one of the remarkable conceptual innovations his group had made. Conventional robots, while undertaking a navigation task, use an approach of 'obstacle avoidance' ie. dodging and evading an obstacle as soon as it is sensed from a distance. The approach he uses is one of classifying 'obstacle impedance' and moving accordingly; so that obstacles are not just dodged but first probed and then possibly pushed aside or even penetrated if their 'impedance' or the constraint they enforce is small enough. A robot operating using obstacle avoidance would stop short on even detecting tall grass in a forest, while a robot using obstacle impedance characterization would know that the grass can well be trampled upon and does not at all represent a barrier. Digressing with an analogy; when we think that a door is locked just because it appears closed and pass it by without even knocking, we are 'obstacle avoiders'. When we are we use a smart approach of classifying 'obstacle impedance', we would nudge even those doors that seem to be closed and find that some are actually ajar and lead to productive encounters. An approach of overcautious withdrawal stalls all advance, and one of alert assessment at every step may reveal paths which were earlier not obvious.

After describing 'obstacle impedance classification' and other innovative features of his mine detection robot, Dr. Nanayakkara described another ancillary area of research,namely, training mongooses to sniff out mines in fields. While traditional sniffer dogs take as long as a year to train and require a human master to accompany it to the life-threatening minefield, mongooses can be trained in as little as two weeks. Who will accompany the mongoose to the field then? Here came the kicker, when he nonchalantly said as if it were the most obvious thing, "We have a robot. We have an animal. Now we tie them together!". While the audience listened astonished, he described how the mongoose 'yoked' to the robot is an effective hybrid system to survey minefields. To prevent the problem of the animal and the robot pulling in different directions, the system incorporates what he calls a 'bargaining mechanism' where the mongoose is trained with Pavlovian methods to respond to signals from the robot when the robot is more likely to be on the right track, and the robot is programmed to respond to feedback from mongoose's movement when it is the mongoose that is one the right track. In a relatively flat field where the robot has superior navigation, this system ensures that the robot pulls and the mongoose obeys. When it is a dense thicket which demands the mongoose's natural skills, it is the robot's turn to obey. Unlike the crude and callous use of live dolphins as minesweepers by the US Navy, the use of mongooses here causes almost no harm to the animal given its naturally gifted nimbleness. Unlike the oft-quoted anecdote about bullocks drawing a motorcar in India (jovially called the Ox-Ford by none other than Mahatma Gandhi) due to the vehicle's malfunction, the mongoose and robot drawing each other is a picture not of subcontinental resource-crunch, but of indigenous resourcefulness. When the state-of-the-art in landmine detection in his country was old-fashioned unreliable metal detectors or more realistically, simply rakes; this pioneering researcher proved how a wild rodent so far seen only as an exotic pet could drive his pet project to unprecedented accomplishment. Indians have a word for this kind of resourcefulness: jugaad, which is vividly described in an article that appeared in the Times of India, which says, "The operative world of jugaad, implying alternatives, substitutes, improvisations and make-dos, is spurred by a native inventiveness steeped in a culture of scarcity and survival."***

The talk had begun of course with a narrative on the horrors of war, and at the end of it, there doubtlessly lingers in mind not just benedictions for Dr. Nanayakkara and his group but an almost urgent prayer for Peace on Earth. A reverie brought to mind these lines from the Bible, which seem so apt and would be so reassuring to those who brave deadly minefields probing for mines with little more than rods, hoping to liberate farmlands and lay the tables of the famished community with the yield of a bountiful, peaceful earth to come.

Yea, though I walk through the valley of the shadow of death, I will fear no evil:
for thou art with me; thy rod and thy staff they comfort me....

Thou preparest a table before me in the presence of mine enemies...

Surely goodness and mercy shall follow me all the days of my life...

-Psalm 23, King James Bible
* Dr. Nanayakkara's University of Moratuwa page

** Dr. Nanayakkara's research and video clips

***Link to the article on 'jugaad' in the Times of India


The Illuminator said...

Swades was an inspiring work of fiction, but Dr Nayanakkara's story is far more inspiring to say the least. However, the technically sound diaspora coming back home for research is a rarity, considering how conducive the environment here is.

There are a few words in colloquial mumbaiya tongue which can't be substituted by an english counterpart, like shot, juggad, ratofying, etc. They just simply expressions.

Seeker of Truth said...

Dealing with these issues simply at the 'buzzword-level' will not help us make much headway. For well-meaning upwardly mobile youngsters, "non-profit" is a dirty word and for career NGO-workers, "corporate social responsibility" is all-too-often just an empty promise and cynicism-trigger. Everyone knows that "micro-funding" is a good idea. When then, can we expect pilot projects of this nature to become the stuff of coursework projects at B-schools; or implemented as public-private partnerships?

Plutohn said...

nice read AI,

This I feel, is similarly inspiring