PhD Oral Exam - Jundong Wu, Mechanical, Industrial and Aerospace Engineering

When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.

Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.

Abstract

Soft robots are robotic systems which are inherently compliant, and can exhibit body deformation in normal operations. This type of systems has unprecedented advantage over rigid-body robots since they can mimic biological systems to perform a series of complicated tasks, work in confined spaces, and interact with the environment much more safely. Usually, the soft robots are composed of subsystems including the actuator, the sensor, the driving electronics, the computation system, and the power source. In these subsystems, the actuator is of great importance. This is because in most situations the actuator works to carry out the operations of the soft robot. It decides the functionalities and physical features of the whole system. Meanwhile, other subsystems work to aid the successful functioning of the actuator. Thus, the study on soft robot actuators, especially the modeling and control of soft robot actuators is the key to soft robot applications.

However, characteristics of soft robot actuators vary greatly due to the usage of different actuator materials. These materials include the variable length tendons, rubbers, smart materials, etc.. Among these different materials, smart material based actuators have the advantage of fast response, light weight, and can respond to various types of external stimuli such as electrical signal, magnetic signal, light, heat, etc.. As a result, smart material based actuators have been studied widely for possible soft robot applications.
Recent years, among smart materials, the liquid crystal elastomer (LCE) starts to catch researchers' attention. LCE is a type of smart material which can deform under the stimulation of light. Unlike conventional actuators, the LCE actuator can be separated from the power source, suggesting a simpler and lighter design, possible for applications that are totally different from conventional electro-driven or magneto-driven actuators. However, just like other smart materials, the deformation characteristics of the LCE actuator exhibits a complicated hysteretic behavior highly dependent on environmental factors, which brings difficulty to the modeling and control. Furthermore, the deformation of the photo-responsive LCE actuator is a multi-step process, resulting greater inaccuracy when compared with conventional smart material based actuators. These are huge challenges that need to be overcome for the modeling and control of the LCE actuator, which is still in its preliminary stage.

This dissertation aims to develop suitable modeling and control strategies for the photo-responsive LCE actuator with the purpose of using it in soft robot applications. Here, by looking into the physical nature of the light-induced deformation of the LCE actuator, it can be concluded that LCE's deformation is inherently the macroscopic shape change resulted from the microscopic phase change of LCE molecules. Based on this deformation mechanism, an experimental platform including a computer, an I/O module, a programmable laser, the LCE actuator, a thermal camera, and a laser distance sensor is established to study the modeling and control of the photo-responsive LCE actuator. Experiments are performed and the results show that the deformation characteristics of the LCE actuator indeed exhibit obvious hysteresis, which is dependent on environmental factors.

Based on the deformation mechanism of LCE, basic modeling scheme and positioning control scheme for the photo-responsive LCE actuator are established. For the modeling of the LCE actuator, the goal is to obtain its temperature-deformation relationship and describe the hysteresis with small errors. Here, the average order parameter is introduced to give a quantitative description of the macroscopic average phase of LCE molecules. Then, the key to obtain the temperature-deformation relationship is to first find the relationship between the temperature of the LCE actuator and the average order parameter, and then find the relationship between the average order parameter and the macroscopic deformation. The overall model is the combination of the above two relationships. Meanwhile, based on the physical nature of the LCE actuator, the positioning control of the photo-responsive LCE actuator is studied. Analysis on the deformation of the LCE actuator from the energy perspective shows that the positioning control of the photo-responsive LCE actuator is a multi-step process, which brings difficulties in control accuracy. To reduce the positioning control errors, a double closed-loop control structure with a feed-forward module is designed for the positioning control of the photo-responsive LCE actuator.

According to the modeling scheme, a basic physical model for the photo-responsive LCE actuator is established. This model aims to develop a quantitative model that reflect the actual physics of the LCE actuator. By assuming that the phase transition of LCE molecules is under dynamic equilibrium at each specific moment, a simple analytical relationship between the temperature and deformation of the LCE actuator can be obtained. For this model, the Landau-de Gennes expansion of free energy for nematic LCEs is utilized to calculate the average order parameter. Then, by assuming that the phase transition process of LCE molecules is under dynamic equilibrium, the relationship between the temperature and the average order parameter is obtained. Meanwhile, thermal dynamic analysis gives the relationship between the average order parameter and the deformation. The above two relationships are then combined together to give the overall model. Model parameters are calculated based on nonlinear least squares method. Experimental results show that this model works to give a good prediction of the deformation characteristics.

Based on the above basic model, an improved model is then established to give a more detailed description on the hysteresis by considering the actual dynamic process of the phase transition of LCE molecules. In order to reflect the actual dynamic process, a small variation of the temperature is considered, and the corresponding number of LCE molecules that undergo phase transition is calculated based on thermal dynamic analysis and a polynomial expansion of the transition rate. As a result, a dynamic equation that gives the temperature-deformation relationship is obtained. To obtain the values of model parameters with efficiency, a two-step parameter identification method based on the differential evolution algorithm and nonlinear least squares method is established. Experiments show that the improved model can describe the hysteretic deformation characteristic of the photo-responsive LCE actuator with high accuracy.

Utilizing the positioning control scheme, controllers are designed for the positioning control of the photo-responsive LCE actuator. For the proposed double-closed loop structure, the inner loop uses a PID controller to control the temperature of the LCE actuator, the parameters of the inner loop controller are tuned using a stimulation-experiment combined method based on the Hammerstein-Wiener model. Meanwhile, the outer loop consists of a PID controller and a feed-forward controller, the feed-forward controller is a numerical inverse model of the simple physical model that is established in the modeling part, and calculates the target temperature for the inner loop based on the positioning control objective. Parameters of the outer loop controller are directly tuned through experiments. Based on the proposed control strategy, experiments with different control targets are carried out to prove that the proposed controller can achieve the positioning control target with high accuracy. Comparison experiments also show that the proposed double closed-loop structure is faster in response, and has smaller control errors than conventional single closed-loop control structure.

In the end, design guidelines for LCE based soft robots are discussed from the application perspective. Designs of a two-legged walking robot and a light-controlled rolling robot based on the photo-responsive LCE actuator are introduced, conclusions are made together with possible working directions for future studies.