next up previous contents
Next: Vision Up: Human Input Channels Previous: Human Input Channels

Input modalities


Obviously, not all of these channels/modalities are of the same interest for . As outlined in the Technical Annex, the project will mainly address the senses of vision, hearing, and the somatic senses. The remaining question is: Do we consider these senses as ``atomic'' or do we also take into account the sub-modalities which have been presented partly in table 2.1 ? We decided to follow the first approach, so in the following paragraphs we will exclude some possible input modalities from the project.

It is possible to distinguish sub-modalities, according to different sensitivities, e.g. with respect to peak frequency or cut-off frequency. In the eye, there are rod and three types of cone receptors in the retina, each with their specific properties. Similarly, in the motor system, there are the bag and chain fibers in the muscle spindles, etc. However, these very low levels of the infrastructure of the perceptual system can be skipped within the context of .

Some of the sensory modalities, however, do not have a cortical representation (sense of balance, chemical senses) or just have a very reduced one (taste) and do not give origin to ``conscious perception'', whatever is the controversial meaning we attribute to this concept; thus we cannot speak, for them, of ``perceptual channels''.

The senses of smell and taste might not be very interesting for (and human information processing in general). This is not due to the fact that ``the corresponding output devices are missing'' but to the fact (i) that taste is not a very useful channel of man-machine interaction and (ii) smell has some practical problems. However, it should be emphasized that the sense of smell has great potentialities, particularly if we consider man-machine interaction with mobile robots: in nature, indeed, odors are not only important from the point of view of ``chemical'' analysis, but also from the navigation point of view, for ``marking'' the territory and setting ``landmarks'' which are of great help in path planning. Also, biological memory is linked to odors, most probably because the phylogenetically oldest systems of territory representation are based on the chemical senses. Spatial memory is probably related to the hippo-campus, which is a cortical area in the immediate neighborhood of the olfactory cortex. Some experiments of robotical path planning, following the gradient of some odor, are reported in the literature and it seems reasonable to consider odor as a creative bidirectional channel of communication between man and a moving machine. Unlike sound, olfactory marks have a physical persistence, like visual traces, but can be invisible themselves and may thus be used in parallel to visible traces.

According to [279], there are many different ways to use odors to create a striking sense of presence:

``The technology of delivering odors is well-developed [343], in trials at Southwest Research Institute. The odors are all Food and Drug Administration approved and delivered at low concentration. The system uses a micro-encapsulation technique that can be dry packaged in cartridges that are safe and easy to handle. Human chemical senses such as taste and smell create particularly salient memories.''
In the context of , however, other input modalities are deemed more relevant: vision, hearing, and the somatic senses. Due to the importance of these senses, they will be considered in more detail in the next sections.

next up previous contents
Next: Vision Up: Human Input Channels Previous: Human Input Channels

Esprit Project 8579/MIAMI (Schomaker et al., '95)
Thu May 18 16:00:17 MET DST 1995