It was only through experimentation, the next step in the scientific method, that answers to these questions were achieved. All experiments must control certain variables; in the experiment regarding distance between the companion robots an human subject, all other factors in the relationship were controlled, and it was determined that most people kept a standard social distance form the robots, while a large minority took up closer positions indicating a less social relationship (Walters et al. 2005). By controlling some variables and measuring a wide array of independent and uncontrolled variables, other researchers determined that there were many factors influencing the perception of companion robots, and that human-like behavior was not actually strongly reacted to (Dautenhan et al. 2005). This same research team, by conducting a series of controlled experiments using different approaches, found that side approaches by robots were generally preferred by most seated human subjects receiving assistance from companion robots, with a certain female subset preferring frontal approaches, and without any regard to specific functions (Dautenhan et al. 2006).

Other research has also been conducted by applying the scientific method to other issues identified in the design of companion robots. By examining specific types of robot-robot and robot-human interactions, one research team was able to identify variables that appear to influence the ways in which companion robots and other self-organizing systems learn (Tani et al. 2005). Further research in more controlled settings is necessary to confirm some of the preliminary findings of this research, and more specific hypotheses need to be developed and tested, but this presents a very different yet highly necessary area of research (Tani et al. 2005).

Other research also examined distinct types of relationships in order to develop more effective and more easily-integrated companion robots, tackling similar issues from a very different angle and examining the specifics of relationship differences rather than changes in robot learning (Ohkubo et al. 2003). This research was more controlled, with many variables kept consistent so as to develop better and more certain understanding regarding the dependent variable in the study, namely effective communication interactions...

2003). Researchers had hypothesized that certain specific functions and styles would be effective in different communication relationships with the companion robot, and the controlled experiments they conducted largely confirmed their hypotheses in this regard (Ohkubo et al. 2003). This is one more way in which the scientific method is helping to make companion robots an effective piece of tomorrow's world.
Conclusion

Whenever it seems that the scientific method is slow or inefficient, it should be remembered that the technologies we take for granted today are the result of centuries of slow developments. It is the combination of many small gains in information that leads to seemingly radical new developments such as computers, cell phones, and soon companion robots. The research detailed above demonstrates quite clearly the necessity and the applicability of the scientific method.