Vision based control of a mobile robot for driving applications
Authors
Grebe, Gregory S.
ORCID
Loading...
Other Contributors
Wen, John T.
Julius, Anak Agung
Fajen, Brett R.
Julius, Anak Agung
Fajen, Brett R.
Issue Date
2017-05
Keywords
Computer Systems engineering
Degree
MS
Terms of Use
Attribution-NonCommercial-NoDerivs 3.0 United States
This electronic version is a licensed copy owned by Rensselaer Polytechnic Institute, Troy, NY. Copyright of original work retained by author.
This electronic version is a licensed copy owned by Rensselaer Polytechnic Institute, Troy, NY. Copyright of original work retained by author.
Full Citation
Abstract
The second controller is based on using the visual angle of a lead vehicle to control and keep a constant following distance. The model is shown to be more numerically stable than a tau based approach, and can be easily tuned to match a desired second order system response. This algorithm is evaluated in both simulation and on a physical robot hardware implementation. Several driving scenarios are presented to showcase the algorithm’s effectiveness, including videos of the results.
Autonomous navigation of a shared roadway is a challenging task that requires an advanced model of the environment. Many autonomous vehicles currently achieve this using an expensive sensor suite based primarily on radar or LiDAR for navigation and control, while relegating vision to more contextual tasks. Nevertheless, vision can be extremely powerful and, as evidenced by human visual perception, provides useful information for the control of a vehicle. In this research, we examine the task of controlling vehicle headway using purely vision based controllers. The main idea is to design robot or vehicle controllers that mimic human visual perception models for the same task. Towards this end, we propose two controllers based on human visual perception models used in vehicle braking and vehicle following scenarios.
The first controller is based on the long studied time-to-contact visual variable, τ (“tau”), as it relates to controlled braking for collision avoidance. In this implementation τ is measured directly from the image features and their derivatives. The control scheme is then based on regulating ̇?? (“tau-dot”), which can be shown to have some very simple relationships to desired deceleration profiles. Ultimately, however, it is shown in simulation that the computations of τ and ̇?? using this method leads to large measurement errors, to the detriment of the overall controller.
Autonomous navigation of a shared roadway is a challenging task that requires an advanced model of the environment. Many autonomous vehicles currently achieve this using an expensive sensor suite based primarily on radar or LiDAR for navigation and control, while relegating vision to more contextual tasks. Nevertheless, vision can be extremely powerful and, as evidenced by human visual perception, provides useful information for the control of a vehicle. In this research, we examine the task of controlling vehicle headway using purely vision based controllers. The main idea is to design robot or vehicle controllers that mimic human visual perception models for the same task. Towards this end, we propose two controllers based on human visual perception models used in vehicle braking and vehicle following scenarios.
The first controller is based on the long studied time-to-contact visual variable, τ (“tau”), as it relates to controlled braking for collision avoidance. In this implementation τ is measured directly from the image features and their derivatives. The control scheme is then based on regulating ̇?? (“tau-dot”), which can be shown to have some very simple relationships to desired deceleration profiles. Ultimately, however, it is shown in simulation that the computations of τ and ̇?? using this method leads to large measurement errors, to the detriment of the overall controller.
Description
May 2017
School of Engineering
School of Engineering
Department
Dept. of Electrical, Computer, and Systems Engineering
Publisher
Rensselaer Polytechnic Institute, Troy, NY
Relationships
Rensselaer Theses and Dissertations Online Collection
Access
CC BY-NC-ND. Users may download and share copies with attribution in accordance with a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License. No commercial use or derivatives are permitted without the explicit approval of the author.