This　paper　develops　a　fuzzy　obstacle　avoidance　system　for　an　elderly-assistant　and　walking-assistant　robot　(EWR)　using　two　ultrasonic　sensors　mounted　at　the　front　of　the　robot.　The　control　architecture　for　a　behavior-based　EWR　is　implemented　in　this　paper.　The　robot　uses　a　reactive　strategy　determined　by　sensory　information　to　interact　with　an　unknown　environment.　The　EWR　uses　tactile　force　to　convey　the　user＇s　intended　motion　action　to　the　robot　via　man-machine　interaction.　For　user＇s　safety　and　avoidance　of　collision,　autonomous　obstacle　avoidance　are　increasingly　used　in　unfamiliar　environments.　For　the　EWR　navigation　in　unpredictable　environment,　fuzzy　behavior　capable　of　making　inferences　is　more　appropriate.　Equipped　with　ultrasonic　sensors,　a　rule-based　fuzzy　controller　with　reactive　behavior　was　implemented　and　tested　on　the　EWR　to　perform　collision-free　navigation.　From　the　simulation　and　experimental　results　obtained,　it　can　be　inferred　that　the　proposed　architecture　provides　an　efficient　and　flexible　solution　for　the　EWR　obstacle　avoidance.　©　2018　IEEE.