In information service systems, the user's demand and service contents are continuously changing depending on economical and social circumstances. Demand-oriented information service system is thus a proposition made to cope with continuously changing conditions of utilization and provision in current information systems. The faded information field (FIF), sustained by push/pull mobile agent technology, is such a distributed architecture that brings adaptability and real-time to the system through a balanced selective replication of the information. When the users' demand changes, the information environment is restructured so that the same response time to services for all unspecified users can be achieved whatever the demand volume. However, once the structure is fixed, it is still necessary to permit to the user's Pull mobile agents to autonomously adapt to the rapidly changing load conditions. In this paper, we propose an autonomous real-time navigation technology in order to permit to the pull mobile agents to decide step-by-step of their navigation behaviors in case of overload in the information environment. Effectiveness of the proposed technique has been confirmed by simulation.