Development of an autonomous robot with artificial vision for obstacle avoidance

Abstract

The implementation of an obstacle-avoiding mobile robot using artificial vision, using the Lego Mindstorms EV3 robotics kit. The robot is configured as a differential mobile robot, through the integration of sensors, actuators and a controller that allow its autonomous movement. The control system and navigation logic were developed in Python, using the Spider environment and the NumPy and Sklearn.Neural_Network libraries, which facilitated the implementation of the neural network by reducing the complexity of the mathematical calculations. For artificial vision, a Pixy camera was incorporated, responsible for detecting objects by recognizing the size and color of the objects and sending the information to the controller, allowing the neural network to make optimal decisions for obstacle avoidance. Concepts of direct and inverse kinematics are addressed to determine the position and speed of the robot, focusing on its geometric configuration. Through tests and adjustments in the programming, the EV3 mobile robot was able to move autonomously in its environment, avoiding obstacles and providing an auditory response on the detected colors. The results demonstrate the effectiveness of the combined use of neural networks and artificial vision in the autonomous navigation of mobile robots.