Author: Zhao, Qingxiang
Title: Design and control of hyper redundant multi-segment pneumatic-driven continuum robot
Advisors: Chu, K. Henry (ME)
Degree: Ph.D.
Year: 2022
Subject: Robots
Robots -- Control systems
Hong Kong Polytechnic University -- Dissertations
Department: Department of Mechanical Engineering
Pages: xx, 129 pages : color illustrations
Language: English
Abstract: Automatic equipment plays an increasingly important role in human daily life, for which can free human beings from tedious and repetitive tasks, can improve effectiveness and can standardize the workflow of implementation. Therefore, the automation level of a country somehow marks the industrialization degree and the development of manufacturing industry. Robot arms are typical representatives in automatic equipment. The wide application of them, such as in manufacturing industry, medical surgery, and search and rescue, have contributed to effectiveness enhancement and cost decrease. Especially with the rapid development of Artificial Intelligence (AI), robot arms are capable of completing complicated tasks requiring little human intervention. This indeed promotes the flexibility in achieving diversified tasks, and the collaboration with human workers or with other robots creates more possibility in replacing traditional working steps. Conventionally, rigid robot arms are common in applications because of the high pose precision, load-carrying performance and robustness. However, due to their inherent structural characteristics including rigid robot body, bulky encoders, motors and transmission mechanisms, rigid robot arms may not be applicable in constrained environments.
In contrast, soft robots that are made of hyper elastic materials exhibit obvious advantages in terms of the safety issue, cost, flexibility, dexterity, and compliance. Through diversified fabrication technologies, many soft robots have been created for specific scenarios, like exploring undersea scenarios and minimally invasive surgery (MIS). Biology has inspired researchers to explore soft robots that are capable of locomotion and manipulation in cluttered environments. Therefore, researchers desired to develop robots with flexible body. Continuum robots whose body is similar to snakes, elephant's trunk and octopus tentacles, have drawn significant research interest in the past two decades. The definition of a continuum robot can be a continuously deformable, infinite DoF and elastic manipulator and they can mitigate some drawbacks of cumbersome rigid robot arms. There are generally three chambers radially distributed inside a single soft segment, where the longitudinal forces along them lead to length difference in each chamber, such that the deformation towards omni direction occurs. The soft material and structure also render continuum robots safe to touch, which further benefits manipulating soft objects, working in narrow space and safe collaboration with human. On the other hand, the high degree of compliance also poses difficulty in against external loads, attracting significant interest of designing robust controllers. Continuum robots are also easy to be disturbed by external forces, making a traditional analysis-based controller infeasible. Researchers in recent years paid significant attention to designing robust controllers for diversified continuum robots.
Although remarkable achievements have been made in continuum robotics, pursuing better actuation mechanisms, finding sensors and achieving high accuracy of control schemes are still hot spots for researchers. As externally configured sensors like stereocameras, can real-time sense the shape and the tip configuration of the continuum robot, this sensing mechanism is not able to work in constrained and occluded scenarios. The first contribution of this thesis is proposing a shape estimation module and a closed-loop controller, forming visualization manipulation system. Strain gauges were employed to act as embedded sensors to sense the robot deformation, and curve fitting algorithms connect the predicted key points by LSTM-MLP NNs. This data-driven method provides a simple solution with the mapping between sensor readings and the true shape configuration. Closed-loop controller scheme integrating accurate feedback contributes to high accuracy of tip configuration control, and the shape reconstruction module can not only sense the shape but the tip position, so that the controller directly obtains the real-time tip position. Data-driven method was also considered to simplify the control architecture, which was based on the inverse of Jacobian matrix. An adaptive step distance mechanism was proposed to adjust the step distance between steps to automatically bypass the obstacles. Specifically, when the tip position is close to the destination, smaller step distance was set, and when the deviation between the planned and the actual is larger, showing obstacles or external forces present, robot system considered bigger step distance to offset the influence.
Additionally, it would reduce complexity of designing controller if the external forces could be accurately obtained, including the acting direction, position and the magnitude. Working in unstructured environment means any external force could present uncertainly, and the shape of the manipulator is jointly determined by the external force and actuation inputs. The second objective of this thesis is to propose a method to estimate the information of the external force. Similarly, proprioceptive mechanism should be pursued to sense the uncertain external force (UEF). Being different from some existing works, we aim at estimating the UEF acting at the circumferential surface of the soft manipulator, and the area where an UEF is likely to present was labelled by a 2D map. Each area was marked by a row position and a column position accordingly. Once detecting the presence of an external force, the column position indirectly reflects the deviation orientation and the row position was related to the deviation degree. Therefore, Hidden Markov Model(HMM) was employed to estimate the column position. Then, to find the corresponding row position and the magnitude, virtual work principle assuming the robot was in balance was utilized. To simplify the calculation process and to pursue accurate estimation, iteration algorithm integrating an optimization factor was designed in finding the magnitude.
Apart from investigating deformation and load-compensation characteristics, this thesis also proposed a novel mechanism that enables omnidirectional bending and continuous rotation simultaneously. The challenge lies achieving rotation along the deformed backbone while maintaining the shape unchanged. To the best of our knowledge, this is the first rotational continuum robot with the capability of omnidirectional bending and rotation. To address the tube twining issue between the actuators and the manipulator, a slip ring was employed to decouple the transmission of the pressurized air. As the newly added revolve joint did not expand the task space, the motion velocity of the manipulator could be enhanced. Based on this, an optimization algorithm was proposed using Genetic Algorithm (GA) to control the tip configuration with the objective of time effectiveness. The continuous rotation along the deformed backbone was achieved by integrating the motion of the base and the deformation. This thesis additionally designed the algorithm to control the rotational behaviour.
Rights: All rights reserved
Access: open access

Files in This Item:
File Description SizeFormat 
6489.pdfFor All Users2.66 MBAdobe PDFView/Open

Copyright Undertaking

As a bona fide Library user, I declare that:

  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.

By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

Show full item record

Please use this identifier to cite or link to this item: