Home
Resource
Modeling and control of physical human-robot interactions with applications to assistive robots for construction workers
PDF
PDF format is widely accepted and good for printing.
Plug-in required
PDF-1(28.31 MB)
Citation & Export
View Usage Statistics
Staff View

Citation & Export
Hide

Simple citation

Chen, Siyu. Modeling and control of physical human-robot interactions with applications to assistive robots for construction workers. Retrieved from https://doi.org/doi:10.7282/t3-bc4b-3v50

Export

Click here for information about Citation Management Tools at Rutgers.

Statistics
Hide

Description

TitleModeling and control of physical human-robot interactions with applications to assistive robots for construction workers
NameChen, Siyu (author); Yi, Jingang (chair); Benaroya, Haym (member); Zou, Qingze (member); Su, Hao (member); Trkov, Mitja (member); Yi, Jingang (member); Rutgers University; School of Graduate Studies
Date Created2022
Other Date2022-01 (degree)
SubjectRobotics, Mechanical engineering, Control systems, Human modeling, Human-robot interaction, Machine learning
Extent151 pages : illustrations
DescriptionThe construction industry is one of the highest-risk private sectors in the US, contributing to 21.6% of all work related fatal injuries in 2019. Work-related musculoskeletal disorders (WMSDs) are among the most prevalent occupational health problems affecting manual work in construction industry. Construction workers (e.g., bricklayers and roofers) are frequently exposed to high physical risk hazard, such as repetitive motions (lifting/lowering), awkward postures and lifting weights, and this work environment potentially causes WMSDs. Additionally, construction workers often carry or handle heavy objects/materials and these manual work can potentially cause injuries and back pains. Construction workers also regularly perform tasks with locomotion and gaits that include walking, repetitive kneeling down and standing up on level or inclined surfaces. For example, roofing leads among construction trades in fatal injuries and has the fourth-highest fatal work injury rate among all occupations. Assist robotic device is deemed as a useful tool to help workers generate the target motions in certain tasks. The main goal of this dissertation is to develop new robotics and automation technologies that can potentially reduce and alleviate WMSDs for construction workers. To achieve the goal, we need to understand the human muscular and postural model, and design robot control system to provide necessary assistance to finish physical tasks in human-robot interactions for construction workers. The topics of this dissertation try to focus on two types of physical human robot interactions in construction: one is the use of robotic manipulator for upper limb assistance and the second for wearable sensor and robot for human walking assistance.

The first part of this dissertation mainly focuses on human-manipulator interaction and presents an assist-as-needed framework to control robotic manipulators. Collaborative robotic manipulators are used to assisting human workers in performing physically intensive tasks. Robotic manipulators augment human upper limb capability (e.g., strength and accuracy), provide necessary assistive forces and help humans perform certain tasks (e.g., move or lift weights and weighted tools). Assist-as-needed control is one of the main concepts to design assistive robotic systems in human-robot interactions. One basic requirement in designing assist-as-needed controllers is the knowledge of human physical strength capacity (e.g., the maximum limb forces) to achieve the desired locomotion. With proper human strength capacity estimation, we design a controller for robotic assistance under changing environments and external disturbances. We take advantage of the muscle synergy model and estimate human physical strength and effort without using electromyography (EMG) sensors. Muscle synergy represents a group of relationship-fixed muscle activities during a particular human limb motion. Human limb motions and their forceful interactions with the environment can be captured as sets of muscle synergy combinations. With these estimations, we design a muscle synergy-based robotic controller to assist human motion and reject unknown disturbance. The control robustness performance and effectiveness of disturbance rejection are demonstrated through extensive experiments that are performed by multiple human subjects.

The second half of this dissertation mainly focuses on human kneeling and walking gaits and the development of a wearable exoskeleton system to assist construction workers during kneeling tasks. It is critical to model and analyze human performance for kneeling, standing and walking on level and sloped surfaces to understand and capture musculoskeletal injury risks of construction workers. We conduct a human postural balance study on standing and kneeling gaits on inclined and high elevated surfaces. To simulate the high elevation, an immersive mixed reality environment is built with an actual inclined roof surface to create somatosensory haptic feedback. We quantify and model postural balance during kneeling and standing through measurements of the center of pressure and sway motion of the upper body under various inclined angles and heights. A mathematical model is also developed to help understand the experimental results and also to provide design guidance for the prevention and mitigation of the fall risk for construction workers. These developments are validated and demonstrated by experiments.

Based on human model during standing and kneeling, a lightweight, wearable sensing and knee assistive device is developed for construction workers during kneeling and standing tasks. Wearable assistive knee exoskeleton provides an enabling tool to reduce and alleviate WMSD risks for construction workers when performing kneeling tasks on level and sloped surfaces. Analysis of kneeling on level and sloped surfaces is performed for single- and double-leg kneeling tasks and extensive experimental results demonstrate the feasibility of using wearable assistive devices to reduce knee contact loads and muscle activations during kneeling tasks. Human subject testing results validated the devices’ effectiveness in reducing muscle activity and knee contact pressure when performing kneeling tasks.

Finally, a real-time gait detection and pose estimation method is developed to monitor workers’ walking gaits on level and sloped surfaces. We propose a single wearable inertial measurement unit (IMU)-based walk gait detection and pose estimation method for human subjects on level and sloped surfaces. The gait detection algorithm is based on a recurrent neural networks and its outcome is then used in the full-body pose estimation. Due to the periodic nature of biped walking gaits, a learned Gaussian process dynamic model (GPDM) and a phase variable that parameterizes the GPMD model are used to predict human arm and leg joint angles in real-time. Extensive experiments of different walking patterns and speeds on level and sloped surfaces are conducted to validate and demonstrate the design.
NotePh.D.
NoteIncludes bibliographical references
Genretheses
Persistent URLhttps://doi.org/doi:10.7282/t3-bc4b-3v50
LanguageEnglish
CollectionSchool of Graduate Studies Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.
Version 8.5.5
Rutgers University Libraries - Copyright ©2024