Method for detecting tumbling state of humanoid robot based on multi-sensor information

The invention discloses a method for detecting a tumbling state of a humanoid robot based on multi-sensor information. The method comprises the step of judging the current stability of the robot by a fuzzy decision-making system by combining the current walking state, posture information and position information of a ZMP (zero moment point) of the robot. Judgment states comprise 9 states: a controllable state, tumbling forwards, tumbling backwards, tumbling leftwards, tumbling rightwards, tumbling left forwards, tumbling left backwards, tumbling right forwards and tumbling right backwards. The method is suitable for judging whether the humanoid robot is in the controllable state when the humanoid robot is about to tumble and obtaining the tumbling direction of the humanoid robot when the robot is uncontrollable, so that a basis is supplied to generation of corresponding protection motion to the humanoid robot. According to the method, the stability of the motion of the robot can be really reflected, and the r