Year: 2018
Type: report
Abstract: In this survey, we list dierent representations for the state and action spaces of a driving policy.We focus on the literature dedicated to decision making rather than perception, and on the context of autonomous driving only.To formulate the autonomous driving problem as a Markovian Decision Process, one must rst describe the state of the system and its encoding in a space S. We list the main state representations found in the literature in sections 1 and 2.Then, one must dene a set of actions A that gives the agent control over the state dynamics, which we explore in section 3.
Author: Leurent, Edouard
Institution: Renault (France)
Language: English
Subfield: Automotive Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2899331514