Service robotic systems are exemplary human-in-the-loop systems in which various forms of interaction between human and the robotic system take place frequently. Due to variability of the user's physical motion and uncertainty of the environment, however, it is usually difficult to model the "interaction" and find appropriate interfaces to handle it. Recently, several soft computing techniques are applied to service robotic systems synergistically for comfortable interaction and safe operation. As a testbed, a new wheelchair-based robotic system is considered in this paper and various interaction/interface technologies are studied. Soft computing techniques are applied for recognizing facial emotional expression, for coordinating bio-signals with robotic motions, and for selecting a suitable image mapping in vision-based control. For demonstration, several experimental results and experiences are presented. (C) 2002 Elsevier Science B.V. All rights reserved.