Skip to Main Content
 

Global Search Box

 
 
 

ETD Abstract Container

Abstract Header

Predicting Desired Temporal Waypoints from Camera and Route Planner Images using End-To-Mid Imitation Learning

Arul Doss, Aravind Chandradoss
http://rave.ohiolink.edu/etdc/view?acc_num=osu1588697131094167

Abstract Details

2020, Master of Science, Ohio State University, Electrical and Computer Engineering.
This study is focused on exploring the possibilities of using camera and route planner images for autonomous driving in an end-to-mid learning fashion. The overall idea is to clone the humans’ driving behavior, in particular, their use of vision for `driving’ and map for `navigating’. The notion is that we humans use our vision to `drive’ and sometimes, we also use a map such as Google/Apple maps to find direction in order to `navigate’. Therefore, in this study, we replicated this notion by using end-to-mid imitation learning. Besides, this work also places emphasis on using minimal and cheaper sensors such as camera and basic map for autonomous driving rather than expensive sensors such Lidar or HD Maps as we humans do not use such sophisticated sensors for driving. Therefore, in this work, we imitated human driving behavior by using camera and route planner images for predicting the desired waypoints and by using a dedicated control to follow those predicted waypoints. The other reason behind this approach is that numerous research [1] have already been conducted in the modular and end-to-end pipeline. Both the techniques were found to be promising, however, even after decades of research, their results were found to be un-generalizable for all road scenarios indicating the need for a better approach. In particular, Waymo researchers [1] have empirically found out that the end-to-end learning approach is not generalizable even with millions of data points. With all that said, this work tries to divide the end-to-end approach and to explore the autonomous driving problem in an end-to-mid fashion by finding a reasonable spot for `mid’ in the end-to-end pipeline. This thesis also includes the work carried out for developing a 3D photo-realistic environment with Lidar and Google Earth images using SLAM, meshing, and StructureFromMotion techniques. The main application of this work is for AV simulations and pre-deployment testing in a simulator environment and the primary focus is to explore the possibilities to generate a simulation environment where one can validate autonomous driving algorithms before deployment and to conduct an experiment which would be unfeasible in real-world settings.
Levent Guvenc (Advisor)
Bilin-Aksun Guvenc (Committee Member)
81 p.
Automotive Engineering; Electrical Engineering

Recommended Citations

Citations

  • Arul Doss, A. C. (2020). Predicting Desired Temporal Waypoints from Camera and Route Planner Images using End-To-Mid Imitation Learning [Master's thesis, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1588697131094167

    APA Style (7th edition)

  • Arul Doss, Aravind Chandradoss. Predicting Desired Temporal Waypoints from Camera and Route Planner Images using End-To-Mid Imitation Learning. 2020. Ohio State University, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1588697131094167.

    MLA Style (8th edition)

  • Arul Doss, Aravind Chandradoss. "Predicting Desired Temporal Waypoints from Camera and Route Planner Images using End-To-Mid Imitation Learning." Master's thesis, Ohio State University, 2020. http://rave.ohiolink.edu/etdc/view?acc_num=osu1588697131094167

    Chicago Manual of Style (17th edition)

osu1588697131094167
563
© 2020, some rights reserved.Creative Commons License Predicting Desired Temporal Waypoints from Camera and Route Planner Images using End-To-Mid Imitation Learning by Aravind Chandradoss Arul Doss is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by The Ohio State University and OhioLINK.