Safe, robust and reliable technologies are needed for service robots with autonomous traveling capabilities to put them into wide use. Especially, robust localization and safe obstacle avoidance in human-shared outdoor environments are mainly discussed in recent research works. 'Tsukuba Challenge' has been held as a robot competition for autonomous traveling ability in real-world by New Technology Foundation and Tsukuba City since 2007. The task of Tsukuba Challenge 2010 was completing the autonomous traveling in the course which includes pedestrian, park, sidewalk, plaza, bridge, stc. The total distance of the course was approximately 1.1 km. Moreover, Tsukuba Challenge 2010 required avoiding capability against other robots for adaptation to the dynamically changeable real-world. We joined Tsukuba Challenge 2010 as a team "Schuderia Frola AIST" and developed the intelligent wheelchair robot 'Marcus'. The newly-developed real-time obstacle avoidance ability was installed on the robot. This method creates and updates occupancy grid map which records obstacle information from laser range finder (LRF) and generates the avoiding route using A* search algorithm. The effectiveness of the method was demonstrated experimentally in the human-shared outdoor environments in Tsukuba Challenge 2010.