This paper proposes a planning and control method of garage parking for car-like robot. The garage parking problem of car-like robot is known as a difficult task because of non-holonomic kinematics' constraints of the car-like robot system and it's narrow work space. Especially, consideration of error in the control of the system and consideration of dynamic environment such as unexpected obstacles make the problem more complicated one and hard to treat. For the garage parking problem, we apply receding horizon control method as a real-time optimization. By applying the method, desirable control is realized for the car-like robot's garage parking problem even if there are unexpected obstacles in the work space and there are errors such as slip of tires. We verify the effectiveness using an experimental car-like robot system. The proposed control method is also applicable for an automatic control system of car-like robot in general cluttered environment.