Date of Award


Publication Type

Doctoral Thesis

Degree Name



Electrical and Computer Engineering

First Advisor

Mehrdad Saif


Collision Prevention, Force Myography, Human Intention Estimation, Human-Robot Cooperation, Wireless Sensor Netweork




Modern industries take advantage of human-robot interaction to facilitate better manufacturing processes, particularly in applications where a human is working in a shared workplace with robots. In manufacturing settings where separation barriers, such as fences, are not used to protect human workers, approaches should be implemented for guaranteeing human safety. Despite existing methods, which define specifications and scenarios for human-robot cooperation in industry, new approaches are needed to provide a safer workplace while enhancing productivity. This thesis provides collision-free techniques for safe human-robot collaboration in an industrial setting. Human-robot interaction in the industry is studied to develop novel solutions and provide a secure and productive industrial environment. Providing a safe distance between a human worker and a manipulating robot, to prevent a collision, is an important subject of this work. This thesis presents a safe workplace by proposing an effective human-tracking method using a sensor network. The proposed technique utilizes a non-linear Kalman filter and Gaussian optimization to reduce the risk of collision between humans and robots. In this regard, selecting the most sensitive sensors to update the Kalman filter’s gain in a noisy environment is crucial. To this end, reliable sensor selection schemes are investigated, and a strategy based on multi-objective optimization is implemented.Finally, safe human-robot cooperation is investigated where humans work close to the robot or directly manipulate it in a shared task. In this case, the human’s hand is the most vulnerable limb and should be protected to achieve safe interaction. In this thesis, force myography (FMG) is used to detect the human hand activities to recognize hand gestures, detect the exerted force by a worker's hand, and predict human intention. This information is then used to control the robot parameters, such as the gripper’s force. Furthermore, a human intention prediction scheme using FMG features and based on recurrent neural network (RNN) topology is proposed, to ensure safety during several industrial collaboration scenarios.The validity of the proposed approaches and the performance of the suggested control techniques are demonstrated through extensive simulation and practical experimentation. The results show that the proposed approaches will reduce the collision risk in human-robot