Monday, April 10, 2006
There have been many problems concerning maids in Singapore over the years. The news has time and again had reports concerning maid abuse, maids dying because of accidents, and also maids committing crimes themselves. A check on google brings up many articles which are just the tip of the iceberg regarding maid problems in Singapore: Who’s fault is it when these problems arise – the maid’s or the employer’s? Personally it doesn’t matter. What matters is the concept of employing another human being from a less well-off part of the world to do our menial and sometimes dangerous tasks (like cleaning windows in precarious positions) for us. It is cruel and inhumane but that is a fact of life. However, if robots were to improve the lives of mankind as a whole, then they will definitely have to be able to take on the role of the domestic servant one day. The Popular Science magazine reports that we are still quite far off from achieving a highly functioning robotic maid, due mostly to the huge problem area of visual perception in robots: The article also mentions that Sebastian Thrun (http://robots.stanford.edu/) of the Stanford AI Lab (http://ai.stanford.edu/projects.html) has been working on this problem but the dream of fully-autonomous robotic maids still seems quite far off. Extensive research has been carried out by him and his team for many years, with huge amounts of documentation written and many robots made, but at the moment having such a robot is still a wish. Another point stated in that article is that Radio-Frequency ID (RFID) is the way to go to make robots able to see and interpret things the way humans do. A check with that leading RFID company (http://thingmagic.com/html/about.htm) mentioned shows that the company is mainly focused on developing RFID products. It’s partners (http://thingmagic.com/html/partner.htm) are also not really focused on robotics, but rather on other applications of RFID. It seems that only one leading member of that company has dealt a bit with robots for pipelines (http://thingmagic.com/html/about-mgmt-yaelmaguire.htm). Nevertheless, the implementation of RFID in the field of robotics is still a relatively new frontier of research and there is growing interest in this area. See the following articles for more information: Despite all the huge obstacles, the owner of ThingMagic believes that a robotic maid will be possible in 20 years time with the help of RFID technology. So what exactly is RFID technology? It is basically an automatic identification method which deals with wireless data transfer between a RFID transceiver (or tag reader) and a RFID tag which is actually a silicon IC chip. This technology is currently fairly widespread, with the ERP gantries & IVUs the most prominent example of this in Singapore. RFID technology is poised to become even more pervasive and powerful, which bodes well for its applications in robotics. To me, there is one apparent big disadvantage of RFID though: everything that is to be recognised has to be pre-tagged beforehand. I don’t think this counts as true machine vision. But it seems to be one school of thought to deal with the problem of poor visual perception in machines or robots. Whatever the outcome, I do hope that in 20 years time there will finally be a decent robot to replace the domestic worker, improving the lives of both would-be employer and would-be employee. Note: The layout seems correct in MS Word and when I click Preview, but on the actual site it's screwed for some reason.