Development of Voice Control Algorithm for Robotic Wheelchair Using MIN and LSTM Models

In this work, we developed and implemented a voice control algorithm to steer smart robotic wheelchairs (SRW) using the neural network technique. This technique used in (NIN) long short-term memory (LSTM) structure integrated with built-in recognition algorithm. An Android Smartphone application was designed configured proposed method. A Wi-Fi hotspot connect software hardware components of system an offline mode. To operate guide SRW, design employing five commands (yes, no, left, right, stop) via Raspberry Pi DC motors. Ten native Arabic speakers trained validated English speech corpus determine method’s overall effectiveness. The method SRW evaluated both indoor outdoor environments order its time response performance. results showed that accuracy rate for reached 98.2% five-voice commends classifying voices accurately. Another interesting finding from real-time test root-mean-square deviation (RMSD) indoor/outdoor maneuvering nodes 2.2*10–5 (for latitude), while longitude coordinates whopping 2.4*10–5 latitude).