Patients with severe disabilities have the problem of being unable to communicate spontaneously in emergency situations. Therefore, there is a need for thought transmitter devices. Gesture recognition method solves the problems of currently developed devices. This research aims to develop a gesture recognition system that uses pose estimation to create an inexpensive system that does not require a brace. Preliminary experiment result shows that the bending and stretching gesture of the right index finger can be recognized with high probability by shooting from a range of 30 to 60 degrees diagonally above the thumb side. In order to shoot the hand from the appropriate position, a camera was attached to the tip of the robot arm, and a control algorithm for it is implemented. Validation experiment result shows that if the system is placed within the range of available placement, it can recognize gestures with a probability of 98.75%.