<![CDATA[The development of robotics world is quite rapid, many robots are created not only to petrify human jobs, even robots deliberately created to replace the role of humans in their daily life. Various types of robots have been created, one of them is humanoid robot types that resembling a human body shapes. The development of humanoid robot in Indonesia is not less rapid with other countries. In the development, walking technique is a major factor in making humanoid robot. Humanoid robots are required to have the ability to walk like a human by balancing the body position when walking so as not to fall. So that required tilt position detection system robot and balancing system when the robot will fall. So to overcome these problems, this study was conducted. Research by implementing complementary filter method on gyroscope and accelerometer sensors to minimize noise or disturbance of sensor output value. So the accuracy of MPU6050 sensor output value at the angle position detector can assist the robot in giving the balancing action. Testing this research is divided into four parts, among others: testing unfiltered sensors, testing alpha value changes on the method, testing robot reaction to input sensor values ​​in realtime and manual. According to the testing, the best alpha is 0.96 with a sampling time of 0.04 seconds. Robot reaction testing of real-time sensor values ​​fails in robot balancing act, because CM-510 is delayed in moving the servo to provide counterbalance motion and can not keep pace with the data processing speed of Raspberry pi 3 B against sensor data with 40 ms sampling time. As for testing robot motion reaction manually with testing gives disturbance from the front, rear, right and left robots can get a balancing reaction in accordance with the expected.]]>
Copyrights © 2017
