Automated Docking of a Small-scale Tractor-trailer Using an Infrastructure-based System
Open Access
- Author:
- Dorris, James Theodore
- Graduate Program:
- Mechanical Engineering
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- July 23, 2014
- Committee Members:
- Sean N Brennan, Thesis Advisor/Co-Advisor
- Keywords:
- automation
tractor-trailer
infrastructure-based
position
orientation
detection
sensors
docking - Abstract:
- This thesis involves the design of a system that utilizes vehicle detection, vehicle control, and vehicle communication to back a tractor-trailer to a desired position and orientation adjacent to a loading dock location. There are currently systems that control a vehicle’s motion; however, many such systems rely on sensors and equipment that is attached to the vehicle itself, are designed for passenger vehicles, or utilize a model-free control approach. In contrast to these existing systems, the goal of this work is to design a system that does not require any sensing or major computing equipment to be located on the vehicle and utilizes a model-based control approach to back a tractor-trailer to a loading dock location. Simulations are carried out, as well as camera-based experimental tests using a 1:14 scale R/C tractor-trailer vehicle. The practicality of using a Light Detection and Ranging (LiDAR) unit for vehicle pose detection is also explored. For guiding the truck to the loading dock location, a state-space path-following controller is used. Also, wireless communication is enabled between a computer workstation and the vehicle, which allows for autonomous guidance of the vehicle. Experiments showed rough agreement with simulated behavior, but discrepancies were found between the two approaches. For example, in experimental tests using a dock-mounted LiDAR unit to detect vehicle pose, some vehicle pose configurations could not be detected. However, it is concluded that LiDAR units could still be useful sensors for vehicle pose detection if implemented using more than one sensor. Overall, this infrastructure-based automated docking system that employs model-based control is demonstrated to be a viable concept.