Abstract
Achieving accurate excavation profiles is a challenging task for excavator operators who typically use the directions of a grade-person to achieve design grades and levels. Allocation of an extra person is costly and requires constant communication between an operator and the grade-person, which reduces productivity. As a result, several machine control technology providers have developed angle sensor packages to monitor the pose of the excavator components in real-time. These systems, however, are expensive and their installation and calibration are costly and time consuming. This paper presents a generic and scalable computer-vision based framework for real-time pose estimation of an excavator's boom and dipper (stick) using low-cost markers installed on the side of the arms. The hardware components of this system are inexpensive and the setup process is quick and straightforward. The system demonstrated promising performance and has been shown to measure boom and dipper angles with an average uncertainty of less than 0.75° in real-time.
Original language | English (US) |
---|---|
Pages (from-to) | 213-229 |
Number of pages | 17 |
Journal | Journal of Information Technology in Construction |
Volume | 20 |
State | Published - Feb 1 2015 |
Keywords
- Computer vision
- Excavation
- Machine control
- Marker-Based recognition
- Pose estimation
ASJC Scopus subject areas
- Civil and Structural Engineering
- Building and Construction
- Computer Science Applications