AI Quadruped Robot Assistant for the Visually Impaired

Legged robots are more versatile than wheeled robots and offer several advantages- for one, they can traverse complex terrain. However, they are expensive and more complex than their wheeled counterparts. This paper proposes a low-cost prototype developed using a spider-type acrylic chassis, Raspberry Pi 4B, Pi camera module, and ESP32 microcontroller. The robot is equipped with an object detector, a face detector and recognizer, a speech recognizer, a Chat-bot and offers different Human-Machine Interface (HMI) methods; suited to assist the visually impaired. For efficient operation and coordination of the programs, the Robot Operating System (ROS) has been used. Hence, users can interact with the robot’s object detector and face recognizer through speech or text. The Raspberry Pi is used for running the programs, and ESP32 is dedicated to controlling the robot.