IIIT Hyderabad Publications
AUTOSTEPS: Autonomous navigation driven by Spatial Temporal Empty Spaces
Report no: IIIT/TH/2016/33
In future, it is possible to have small sized autonomous vehicles traveling along with larger vehicles on the same roads. Autonomous vehicles are no longer in the realm of science fiction, thanks to the artificial intelligence and integration of numerous auto motive technologies that have matured over many years. Autonomous vehicles could save lives, chase criminals, reduce accidents, and help the environment by reducing fuel demands and pollution. One such example can be police motorcycles. Such autonomous vehicles has to both keep track of the other vehicles and also drive his/her vehicle safely and fast. In this paper, we present AUTOSTEPS, a multi-agent simulation environment where autonomous small vehicles perceive and learn about the environment for fast navigation to reach their goal. The goal for a agent is defined as the destination he wants to reach. The key idea is the notion of spatial temporal spaces into which the small vehicles continue to move till they reach their goal. These spaces are present on the road between vehicles and may change with time and position of vehicles. We built a simulator and ran experiments to show that our strategies for AUTOSTEPS ensures fast and safer navigation for small autonomous vehicles.
Full thesis: pdf
Centre for Data Engineering
Copyright © 2009 - IIIT Hyderabad. All Rights Reserved.