Control of Redundant Manipulators by Fuzzy Linguistic Commands Koliya Pulasinghe 1 , Keigo Watanabe 2 , Kiyotaka Izumi 2 , Kazuo Kiguchi 2 1 Dept. of Production & Control Tech., Graduate School of Sci. & Eng., Saga University, 1-Honjomachi, Saga, 840-8502, Japan 2 Dept. of Advanced Systems Control Eng., Graduate School of Sci. & Eng., Saga University, 1-Honjomachi, Saga, 840-8502, Japan koliya@ieee.org Abstract: This paper presents a method of controlling redundant manipulator by spoken lan- guage commands consisting fuzzy linguistic information. The present system introduces the fuzzy- neuro control paradigm to the contemporary speech controlled robotic systems, which are based on on-off control paradigm. The system is sensitive to the action activation commands, action modification commands, and action repetition commands of the human-robot conversation carried out by practical dialogues. Credibility of the proposed system is experimentally proved by control- ling a manipulator with seven degrees-of-freedom by fuzzy linguistic information enriched spoken language commands to perform an assembling task. Keywords: Natural language commands, Fuzzy-neuro controller, Spoken language based robotic control. 1. Introduction Robotic presence in the human environment has made the researchers busy with investigating a candidate com- munication medium to build the human-robot relation- ship. Following list substantiates the candidacy of the spoken language as the human-robot communication medium and explores the possible areas that the spoken language based robotic systems can be implemented. • The world’s elderly population is accelerated by de- clining fertility rates and steady development in life expectancy. Almost all developed countries as well as developing countries encounter the problems of adopting elderly citizens in near future. Govern- ments have to pay much attention for nursing and aiding services for elderly people. According to the U.S. Census Bureau forecasts, in 2030 the percent- age of European population that is more than 65 years old is 24.3 percent and that of North Amer- ica is 20.3 percent 1) . Under these circumstances, elderly people have to live alone without proper care of their children. Under these circumstances nursing and aiding robots have taken much atten- tion. But, elderly people cannot operate precisely a joystick or similar conventional control equipment as the input medium when they are communicat- ing with human friendly robots. Therefore, spoken language is a better alternative for them in com- municating with human friendly robots, which are employed for nursing and aiding. • People suffering from quadriplegia (paralysis in all four limbs) have no other means than the spoken language that they can utilize to communicate with human friendly robots. Spoken interfaces not only improve the quality of these severely injured peo- ple’s life but also add their contribution to the ef- fective workforce of a country. • A natural language based interface can add an ad- ditional dimension to the work carried out by a per- son, whose eyes and hands are busy with the job. As an example, while driving a car or while carrying out a surgery, people can control the human friendly robots around them by the spoken language. • Tele-operation is another field, where we can ap- ply spoken language interfaces to control robots. Since we already have well established phone net- work, spoken language can be utilized for control- ling robots in remote environments, which will in- crease the human potential. This is very cost ef- fective because speech signals consume low band- width compared to the other signals. In addition to that, natural language based interfaces can be im- plemented for the entertainment robots, which are working remotely in home environments. • Spoken language interface is a better alternative if there are severe space limitations in implement- ing keypads and visual display units on the human friendly robots. It can be implemented with mini- mal power and space requirements, which are ideal for portable or wearable robots. In early days of the speech controlled machines, ma- chine functions are activated by comparing the input user utterance with a stored template as in voice con- trolled wheelchair developed by Mazo et al. 2) . Each com- mand is restricted to one or two words and has an asso- ciated function 2, 3) . To control the machine, user must SICE Annual Conference in Fukui, August 4-6, 2003 Fukui University, Japan PR0001/03/0000-2819 ¥400 © 2003 SICE -2819-