Fixed Obstacle Detection for Autonomous Vehicle

Obstacle detection in autonomous vehicles is mandatory for maintaining the safety of the driver and the vehicle during the trip to the required destination. Currently, vehicles are integrated with alert system to help the driver to drive the vehicle safely through the path. However, autonomous vehicles must detect the obstacles by itself and start navigating through the surrounding objects safely. Various systems were introduced for obstacles detection using different sensor types such as lidar, camera and radar. Raspberry pi camera is used to detect the obstacles ahead of the vehicle during the trip. The obstacles which the vehicle should deal with is divided into two types: fixed obstacles (e.g. Stop signs and traffic lights) and sudden obstacles. The work presented is focusing on fixed obstacles detection using a monocular camera and Raspberry Pi. Python and OpenCv's machine learning libraries are used for image processing using Haar feature-based cascade classifiers method, which detects the objects with high accuracy and low computational time. Moreover, a software made using deep learning technique to identify the obstacle. The output of the two obstacle detection techniques is compared in terms of speed and accuracy. Finally, an I2C (Inter-integrated Circuit protocol) is used for communication between the Raspberry Pi and the main controller of the targeted vehicle, to take the required decision based on the detection result.