Four-wheeled electric vehicles effectively assist users, such as the elderly, in their movement but the dynamic characteristic of these vehicles is too complex to move abeam, making them difficult inexperienced users to drive. To predict the future state of dynamic obstacles and how to elude such obstacle safely, we used a predictive fuzzy controller (evaluates the future states which several kinds of operation candidates were done and chooses the best one) that modeling human algorithms in the system. Human predict the future states of dynamic obstacles and choose a driving operation to safely elude them. To do so, we incorporated expert knowledge of safe driving into this controller. In this paper, we propose and develop the intelligent soft driving system by a controller that eludes dynamic obstacles safely, and confirmed its effectiveness in experiments.