Free Space Detection for Autonomous Vehicles in Indian Driving Scenarios

This paper aims to address the problem of free space detection and safe path estimation for an autonomous vehicle by performing semantic segmentation on the road image data, captured from the front facing camera of a vehicle. The potential for accidents and economic losses can be significantly reduced by analyzing the road surface for obstacles and identifying safe directions for the vehicle to navigate. This involves the use of advanced algorithms for detecting and classifying objects on the image plane, and then determining the safest direction for the vehicle to proceed. The proposed work has significant implications for the development of autonomous driving technology, and its potential to revolutionize transportation by improving safety and reducing traffic congestion. This paper focuses on detecting free space for an autonomous vehicle by utilizing different state-of-the-art deep learning architectures, namely SegNet and UNet. The goal is to locate a path that is free of obstacles and can be used for safe autonomous navigation. The proposed method provides accurate and efficient detection of free road surfaces and obstacles, making it a valuable tool for autonomous driving technology. Moreover, this paper mainly focuses on complex driving scenarios of the Indian roads.