The Human-machine Interface

ELECTRONICSTALK The Independent Weekly Product Information Guide for Electronics Design, Development and Manufacturing Engineers Laurence Marchini, Editor Electronicstalk, writes: The human-machine interface has come a long way since those lessenlightened times when we used to call it the man-machine interface.Along with many different innovations in display technology, the inputside of the equation continues its slow evolution. But many of the basic concepts remain the same: the QWERTY keyboard layout remains,and the mouse (or equivalent) has the same basic function as thatpopularised on the original Apple computers in the 1980s. There have been several attempts to redesign the basic keyboard, but we are still left with a design that owes its layout to the need toavoid key clashes on mechanical typewriters. Nonetheless, the QWERTYlayout - once familiarised - appears to mesh pretty well with thehuman brain, as anyone who can remember the typing pools of the lastcentury will recall. However, regardless of whether the QWERTYkeyboard is traditional electromechanical, membrane or made up oftouch sensors, the fundamental problem remains one of size: each keyhas to be large enough to accommodate human fingers - including thesomewhat un-co-ordinated digits of ham-fisted scribes such as myself. So what of alternative input methods? Well, the past two decades haveseen some laudable attempts to break away from the size limitations ofthe keyboard.In the early 1990s, speech recognition was touted widely as thesolution to all our woes. The arrival of low-cost DSP devices wouldenable a new class of algorithms that could (with suitable training)do away with the keyboard altogether. I vividly remember visiting oneof Europe's foremost language laboratories where speakers of everylanguage and dialect were busy recording every possible syllable andphoneme in an attempt to build the ultimate voice-recognitionalgorithm.And while embedded applications such as voice control of hands-freephones are commonplace, the lack of an all-encompassing speech to textfunction only serves to prove what a diverse thing the human voice canbe, even when we are supposedly talking the same language. After all,who could expect a machine made in California to understand the voiceof a Cornishman.Handwriting recognition was also mooted as an alternative during the1990s. Yet one by one I saw my colleagues discard their styli and goback to their keyboards. It's ironic that one of the applicationscited for the technology in those days was in keeping medical records.And if anyone could crack the code of the average medicalpractitioner's handwriting, they would have found the proverbial HolyGrail!The keyboard may remain king for text input, but today's contendersfor replacement HMIs focus more on navigating around the screen.Vision sensors can be trained to recognise gestures and eye movements. There are even further advances in physical switching devices - as youwill see below, C+K Components has devised a single navigation devicecapable of no fewer than 14 different functions.However, I didn't get where I am today without reading an awful lot ofscience fiction and fantasy in my youth, and so I am particularlyintrigued by the impending commercial release of the world's firstbrain-interface headset by Australian/US crossover company Emotiv. TheEPOC headset incorporates 16 sensors that measure electrical impulsesfrom the brain - fortunately without recourse to the invasive needleprobes favoured by my favourite authors! Not surprisingly, the initial market for EPOC will be in the gamesmarket. But there are innumerable other tasks that could benefit froma 'direct' interface between the human and the machine. Best wishes Laurence Marchini,