Abstract—This paper presents a trajectory generation process of a humanoid robot (HR) to restore its balance from a moving floor. The experiment applies a 1kg iron weight hits to the slidable floor(cart) while a robot is standing on it. The impact produces a 2.97N external force to the cart and thus makes the robot unstable and later falling down. The objective of this work is to determine a balance recovery motion or trajectory, using an inverse kinematics model together with robot’s joints control strategies. Two joints control strategies are implemented i.e. ankle and ankle-hip. The performance of the trajectories is demonstrated through the jerk. This can be analyzed via the HR’s joint angle profile using the cubic spline technique. These profiles are then applied to the real humanoid robot. The satisfied results are achieved for both strategies. However, the ankle-hip strategy gives a shorter recovery period than the ankle strategy.