Self Training Autonomous Driving Agent

Shashank Kotyan, Danilo Vasconcellos Vargas, U. Venkanna

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

Intrinsically, driving is a Markov Decision Process which suits well the reinforcement learning paradigm. In this paper, we propose a novel agent which learns to drive a vehicle without any human assistance. We use the concept of reinforcement learning and evolutionary strategies to train our agent in a 2D simulation environment. Our model's architecture goes beyond the World Model's by introducing difference images in the autoencoder. This novel involvement of difference images in the auto-encoder gives a better representation of the latent space concerning the motion of the vehicle and helps an autonomous agent to learn more efficiently how to drive a vehicle. Results show that our method requires fewer (96% less) total agents, (87.5% less) agents per generations, (70% less) generations and (90% less) rollouts than the original architecture while achieving the same accuracy of the original.

Original languageEnglish
Title of host publication2019 58th Annual Conference of the Society of Instrument and Control Engineers of Japan, SICE 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1456-1461
Number of pages6
ISBN (Electronic)9784907764678
DOIs
Publication statusPublished - Sep 2019
Event58th Annual Conference of the Society of Instrument and Control Engineers of Japan, SICE 2019 - Hiroshima, Japan
Duration: Sep 10 2019Sep 13 2019

Publication series

Name2019 58th Annual Conference of the Society of Instrument and Control Engineers of Japan, SICE 2019

Conference

Conference58th Annual Conference of the Society of Instrument and Control Engineers of Japan, SICE 2019
CountryJapan
CityHiroshima
Period9/10/199/13/19

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Industrial and Manufacturing Engineering
  • Safety, Risk, Reliability and Quality
  • Control and Optimization
  • Instrumentation

Fingerprint Dive into the research topics of 'Self Training Autonomous Driving Agent'. Together they form a unique fingerprint.

Cite this