Obstacle Detection and Avoidance Using Sensors and Deep Learning Models

Abstract

To study various technologies used in order to solve the problem of obstacle detection and avoidance and develop a system capable of efficiently detecting and avoiding obstacles and selecting a path that is free from congestion. This system integrates smart applications with computer vision, Arduino and sensor technology obstacle detection. It autonomouslyidentifies obstacles and avoids them by using the ultrasonic sensors. For the CNN implementation for Level 2 Autonomous Vehicle, a CNNapproach with mapping of camera input pixels to steering commandsand testing on CARLA open-source driving simulator, utilizing an ultrasonic sensor and an RGBD camera for obstacle detection and real- time position monitoring at 10Hz. It incorporates the usage of RRT-Connect algorithm for path planning, employing Arduino Mega and Raspberry Pi for motor control and processing. Perception Module Development for Obstacle Detection in Farm Equipment involves usage of multiple sensors for obstacle detection and motion control in farm equipment with a combination of camera, mini Lidarmodule, and ultrasonic sensors for accurate obstacle detection and distance measurement.