Innovations

Innovations

Keywords: Automation, Cybernetics, Research

1. New AI system

At its research centre in New York, USA, IBM has developed a new AI system that has great potential for a wide range of applications. The system is called Veggie Vision and it uses artificial intelligence software and a modified photocopier to perform its tasks. The idea of the system's developers is to produce an "intelligent" device that can, initially at least, be capable of performing some of the tasks that are necessary in shops and supermarkets. The system could soon perform the task of recognising and weighing fruit and vegetable produce automatically after they have been selected by a customer. The system is said to be so sensitive that it is able to identify and weigh a particular variety of apple or lettuce. One of the senior officials of IBM who is responsible for retail industries for the corporation says that trials have already been carried out. He says that they have perfected this technology in the laboratory but that the real world is completely different so that it has to be tried out in the targeted retail outlets both in the USA and abroad.

The new device, Veggie Vision, will, it is claimed, replace the current barcode scanners. It is able to take a 3-D scan of an item so that the system is able to build up a profile of it. It can assess its size, shape, texture and even its weight. With this observed detail it can compare the data with that stored in a database of the produce, for example, stored in the particular food outlet. This means that a customer's purchase of produce could be matched in say, a supermarket, and the cashier could also see a picture of the item for verification and for pricing. Although the currently developed system is said to be able to differentiate between different types of apple and tomato it has some difficulty in recognising greens. The IBM developers say that:

We are getting 98 percent accuracy in the laboratory, but greens are proving a problem, they all look so similar. Something like lettuce looks much like another lettuce. However, we are getting there, and the system actually learns as it goes along, so over time accuracy will rise.

This is where the implementors of the system need to improvise in the best traditions of AI. If Veggie Vision is uncertain of the vegetable it needs to identify, it will show the cashier its best guesses and ask them to choose the correct one. It is then that the system updates its database to learn how to recognise vegetables that have proved not easy to recognise.

IBM, however, has made the point that the system is experimental and will not be available for at least five years unless there is breakthrough in technology. At present we are told that the system relies on a large sapphire crystal to form its lens. Not only is it expensive, but it is also regarded as too delicate for everyday use. Even so the developers say that costs are bound to fall when supermarkets worldwide take it on and purchase in bulk.

The philosophy behind this system is that there is a need for such a system because the amount and variety of fruit and vegetables available in our supermarkets has grown enormously. The public seems to demand a modern technological system to deal with the selection and purchase of fresh foods and IBM is more than ready to supply the supermarket owners with the latest"intelligent" devices.

Innovative medical information systems

Researchers at the Massachusetts Institute of Technology (MIT), USA, have announced a new innovative communications system which will, for example, allow medical doctors to receive and transmit medical information from a patient's bedside at any location. It would mean that in its medical usage the system would allow a general practitioners (GP) to use a mobile phone to take scans and also to view a patient's medical records. The MIT team say that the system,called The Handy 21, will be able to transmit ultra-sound and electrocardiograph signals from the patient's bedside and also receive information about the patient from medical records. It also will work as a cellular telephone, pager,radio, digital television and internet connector. The developers do say,however, that despite this range of functions it will still be no bigger than an ordinary mobile phone because it has a shared processor. The researchers tell us that the next generation. of wireless computer networks will be based on the same technology.

The MIT team of researchers is headed by Professor John Guttag, who is quoted as believing that:

Any activity like communications involves processing signals, whether the device is a TV scanner, radio or ultra-sound scanner. Historically the processing of these signals has been done on dedicated, special-purpose devices. We decided to take it to the next step and see if we could eliminate the special purpose hardware entirely and do all the processing with just off-the-shelf PC technology. A sensor has only to be added to the Handy 21 to allow it to take different scans -- such as magnetic resonance imaging and ultrasound. Users,such as doctors, would plug a sensor into a port on the device and be able to take scans instantly. If a doctor, for example, needed a second opinion, then scans could be sent to another doctor. It will be like a compressor in the garage that you plug your tools into to inflate your car tyres and do all different sorts of different jobs. Nobody has a separate compressor for each job.

The medical application provides the best example of the use of this technology. It would mean that doctors could take all they need to a patient's bedside, wherever that may be, whether in hospital, at home or at any other location, and transmit the data to a base unit at their local hospital for any further analysis. A doctor could then carry out diagnostic tests on patients at any location and send the information to any destination and be able to receive information in return within minutes.

Although this technology has been publicised by the MIT developers in the context of healthcare, the same technology will obviously transfer to numerous other application areas. In future it would appear that more and more professionals at work in their various locations will use this system as a comprehensive information processing tool. Communication systems that work on this type of technology illustrate the goals that cyberneticians and systemists have been attempting to reach for many decades.

3. New optical reading devices

A report from the UK's Optoelectronic Systems Integration (OSI) programme gives details of a sponsored project where:

Practical opportunities for shrinking the size and cutting the cost of optical bar code scanners and read heads are being opened up.

We are told that substantial progress towards developing miniature bar-code reading systems has been made on the OSI's three year Microengineered Optical Scanners and Read Heads (MORES) project, which was started at the Imperial Colleges's Department of Electrical and Electronic Engineering in 1998. The principal investigator for the MORES' project, Professor Richard Syms, believes that:

There are formidable problems in designing optical systems to overcome the loss of signal that follows from size reduction, as well as in developing appropriate integrated fabrication processes. However, if our techniques are successful they could create a wide range of chip-based optical reading devices.

Such new microengineering techniques could create new optical reading devices. The report continues by informing us that:

MORES is initially focusing on "remote" scanners which operate at a distance from objects, typified by supermarket bar-code scanners. Currently,these cost about »1,000 each and the smallest read head is a one-inch cube. By miniaturising and integrating optical and mechanical functions in the read head,scanners could become more mobile and cheaper. This could assist, for example,in providing shoppers with their own scanners. The MORES approach combines silica-based waveguide optics with moving mechanical components, such as scanning mirrors formed by silicon microengineering.

The final system being developed will use a combination of "pre-lens" and"post -lens" deflection methods to obtain a 2-D multi-line pattern of the object points being scanned.

Pre-lens deflection is being investigated through a prototype consisting of a moving wave guide device, built from a simple etched baseplate and resonant fibre scanner actuated piezoelectrically. In this prototype, deflection of the waveguide effectively moves an object point across a lens, so that the image point tracks across the bar-code. Professor Syms emphasises that:

Scan lines of appropriate length have been demonstrated and signals have been recovered from real product bar-codes on a wide range of backing materials.

The post-lens avenue is being explored through novel self-assembly techniques that have been developed at Imperial College, London, on an earlier project. The report indicates that:

These construct 3-D micro-optomechanical components by first fabricating flat parts on a wafer consisting of a thin silicon layer separated from the underlying substrate by an oxide film. After patterning and etching the silicon and removing the underlying oxide, the parts are automatically rotated out-of-plane by the surface tension forces obtained by melting pads of photoresist material. A simple latch mechanism limits the rotation to 45°Mirrors in this system have a base width of one millimetre, which suggests that components of realistic size may be built by this method.

The research is being supported by the UK's Engineering and Physical Sciences Research Council (EPSRC) which is responsible for a successful and wide-ranging programme in information technology and computer science. Readers can obtain more information from the EPSRC: contact: E-mail: michelle.kennedy@epsrc.ac.ukResearch contact concerning this reported project can be made through: www.ee.ic.ac.uk/hp/staff/sjwill/optical/OptSemiDev.html- Optical and Imperial College Semiconductor Devices Section.

4. Producing the next generation silicon chips

The research team at the Dundee University's Advanced Materials Centre,Scotland, UK have developed new ways of producing next generation silicon chips. Their project investigated the use of a newly synthesised range of organometallic compounds to make photomasks used in the production of high density chips.

The new processes they have developed are for the manufacture of the high-density integrated circuits which, they say, will be required by industry in the future. Professor James Cairns, the project's principal investigator,reports that:

Not only did we demonstrate the effectiveness of this new photomask technology,. but during the research we discovered that the compounds could be decomposed directly to metal under ultra-violet (UV) irradiation. This opens up exciting prospects for yielding high-density metal interconnects in a more direct way than previously feasible.

Dundee's new organometallic compounds, it is claimed, can meet the stringent requirements of the mask writing process inherent in the production operations. This is because the new compounds can be deposited from the vapour phase on to fragile X-ray transparent membranes and can be converted to sub-micron metal patterns using a focused electron beam. At the UK's Rutherford-Appleton Laboratory the X-ray source has been used to demonstrate that this technology can produce X-ray masks with metal line widths down to less than 60 nanometres. The university says that worldwide patents have been filed to protect the technology, and opportunities for exploiting these innovations commercially are being explored.

Further details from the Dundee Microfabrication and Electronic Materials Research Group - www.dundee.ac.uk/PhysicsEngineering/mcrores.htm