This paper presents the development and real-time deployment of an EMG-based gesture recognition system using surface electromyographic (sEMG) signals for robotic arm control. The system classifies four forearm gestures, hand extension, hand flexion, thumbs up, and fist, from sEMG signals recorded from four key muscles. Preprocessing is performed by bandpass and median filtering, followed by the extraction of five time-domain features. A k-Nearest Neighbors (k-NN) classifier is trained on a dataset collected from 35 participants; the model is trained and implemented on a Raspberry Pi embedded platform for real-time inference and gesture-to-motion mapping using the Arduino Braccio++ robotic arm. Each gesture corresponds to a distinct pose in the robotic manipulator, which is made visible to represent the user’s intention. This work reports findings from a pilot study aimed at validating the system’s technical feasibility and establishing baseline performance. This work demonstrates the feasibility of real-time EMG-based gesture control, highlighting its potential for applications in assistive and wearable robotics.