Opportunities and challenges of machine vision for navigation and obstacle detection in autonomous ground vehicles

Abstract

Autonomous vehicles are the future of transport and also it is expected to become a fully-fledged reality within a decade. All the major giants in the automotive industry are hard pressing their transition from conventional vehicle to autonomous vehicles. The state of Karnataka, for instance, had approximately 205,200 registered taxis higher than Madhya Pradesh 174,900 registered cabs from 2014 to 2015. This presents a great deal of opportunities for autonomous cars and need for technologies. Autonomous cars reduces the accidents rate, stress free parking, saves time, reduces traffic congestion, improve fuel economy etc. It is so sophisticated to the level of easy prediction of physical objects, behavioural elements such as driving speed limits and driving rules between the physical world and its map. Autonomous vehicle have grown to an extent of updating its own information and also based on the cloud, benefitting the systems of all other cars on the network. Machine vision is the most crucial aspect which gives the autonomous vehicles the knowledge of its surrounding. This paper deals with the different approaches of machine vision that helps the vehicle in lane and obstacle detections. Few methods of obstacle detection like Single Object Detection and tracking (SODT) and Multiple Object Detection and tracking (MODT) are compared and contrasted in this paper. Despite the enormous advantages, there are still some challenges of autonomous which needs to be addressed. The challenges that the field will face, especially in relevance with India, along with the suggestion for improvement is also discussed.