Multi-Dimensional Localization for Autonomous Vehicle in Urban Environment

Abstract

The objective of this research is to solve the multi-dimensional localization problem for autonomous vehicle in an urban environment, including metric localization, semantic localization, and activity localization. Localization finds the position of a vehicle in its environment, which is one fundamental function for vehicle autonomy. While metric localization is a traditional research topic which estimates robot spatial position, this proposal extends the meaning of localization to the semantic level and activity level which tries to understand nearby environments and people’s motion flow. The map building problem is solved simultaneously together with the localization problem. A general framework is proposed. The idea is applied to an instrumented autonomous vehicle, and preliminary results have been achieved.