OREGON STATE UNIVERSITY
Open search box
Simulation and Analysis of Construction Operations |
Real-time Monitoring and Control of Construction Sites |
Construction Safety |
Building Automation |
Simulation of Construction Operations
The complex nature of construction operations necessitates innovative means for their analysis and optimization. Discrete Event Simulation (DES) has been as the most appropriate means of modeling construction operations on the basis of its flexibility, generality, and modeling approach. Dr. Louis has created a standalone software application called jStrobe, which implements the STROBOSCOPE modeling language for DES (Martinez 1996). In addition to implementing DES modeling and analysis capabilities, jStrobe also enables operations visualization and communication with external construction resources for real-world monitoring and control. Please contact Dr. Louis for a free copy of jStrobe for educational purposes.
Integrating Discrete Event Simulation and Deep Reinforcement Learning for Optimizing Construction Planning and Control
Students involved: Vivswan Shitole
Collaboration with: Prasad Tadepalli (Professor in EECS, OSU)
There is a growing interest in utilizing real-time data facilitated by the ubiquitous connectivity and sensors on equipment to inform decision-makers and optimize operations. This research combines the modeling power of discrete event simulation (DES) with deep reinforcement learning based on Markov decision processes to determine the optimal policy for construction operations with a given set of constraints. Furthermore, since the data itself could depend and vary based on the policy used, the policy is continually fine-tuned and optimized during the course of the operation. The developed methodology is illustrated using the case study of a hypothetical earthmoving operation.
Operation-level Monitoring of Worksites
Present day construction equipment are loaded with a wide variety of sensors that can give you information about its location, fuel used, tire-pressure, etc. The challenge for the construction manager lies in converting all of this equipment-centric data into operation-specific insights that can positively impact their bottom-line. This project showcases how sensor information from individual equipment can be synthesized and transmitted to a discrete-event model of the operation in jStrobe to deliver real-time performance measures to the manager. The developed framework was demonstrated using a virtual construction site developed in the Virtual Robot Experimentation Platform (V-REP).
Operation-level Control for Construction Automation
The growing capabilities of autonomous vehicles and robots herald a new age for construction wherein dangerous, repetitive, and difficult tasks could be performed without humans. Despite the advances made in automating individual pieces of equipment, there remains the challenge of having them coordinate and work together as a fleet to accomplish the overall objectives of the construction operation. This project utilizes the communication capabilities of jStrobe along with its DES modeling features to coordinate a team of trucks and excavators to work autonomously in accordance with a predefined set of operational constraints. The case studies used to demonstrate the capabilities of the developed framework were implemented virtually in V-REP and by using scaled-down model equipment.
Activity Recognition for Construction Equipment
Students involved: Khandakar Rashid
The ability to automatically classify activities performed by various equipment in real-time can assist project managers in reliable decision-making and project control. Such an endeavor requires the identification of individual sequential work-motions (e.g.: excavator swinging empty) performed by equipment, which then combine to a specific activity (e.g.: excavator loading truck). Towards this end, this project developed an automated activity recognition framework for equipment using inertial measurement units (IMU) attached to the equipment’s implements (such as boom, stick, bucket for excavator) that are currently used to determine the position of the cutting edge for automated machine guidance and control purposes.
Internet of Things for Worker Safety
Students involved: Kelsey Chan
Construction has hazardous work environments due to workers in close proximity to other construction entities resulting in an increased risk of safety incidents. Previous research studies concentrated on issuing proximity warnings to prevent these hazardous interactions. However, these previously established systems typically generate false positive alarms that eventually cause workers to ignore these warnings. This project develops a system to minimize the occurrence of false proximity alarms on construction worksites for proximity and visibility related hazards by tracking the field of view of the worker in addition to their position.
Safety Vs. Mobility on Construction Workzones
Students involved: Kenny Fiawoyife
Electrical Appliance Control for Smart Home Environments
Students involved: Khandakar Rashid, Kenny Fiawoyife
User interaction with connected devices in smart homes suffer from numerous disadvantages such as complex and cumbersome user interfaces, unreliable voice and gesture interfaces, and the requirement of prior knowledge of the environment. This research developed an intuitive point-and-click framework to control electrical fixtures in a smart built environment. Apart from the provision of a novel user-interaction mechanism for electrical fixtures, the proposed framework provides a means of extending the utility of building information models (BIM) in the operation phase of a facility and provides a means for its integration with IoT.