The objective of the work described in this thesis is the development of a configuration estimation scheme for quasi soft-bodied robots, with the end goal being accurate soft-robot proprioception to enable robotic manipulation of unknown external loads. The first chapter introduces the problem and state-of-the-art modeling methods. The second chapter presents work we did building on previous research on a novel sensing scheme for soft-bodied robotic configuration estimation. The third chapter discusses the development of a geometric shape-sensing model based on overlapping tendon-length measurements. This model is geometrically exact for a body composed of two constant curvature segments in bending. The next chapter discusses our implemention of this model on a real world system and tested in two and three dimensions. We estimated the shape of a 215 mm long robotic segment with less then 3 mm of median error for a set of 50 configurations causing motion in two dimensions and approximately 27 mm of median translational error in three dimensions. The final chapter draws conclusions and proposes future work to allow for full robotic proprioception.

This thesis focuses on the development and extension of nonlinear evolutionary model predictive control (NEMPC), a control algorithm previously developed by Phil Hyatt of the BYU RaD Lab. While this controller and its variants are applicable to any high degree-of-freedom (DoF) robotic system, particular emphasis is given in this thesis to control of a soft robot continuum joint. First, speed improvements are presented for NEMPC. Second, a Python package is presented as a companion to NEMPC, as a method of establishing a common interface for dynamic simulators and approximating each system by a deep neural network (DNN). Third, a method of training a DNN approximation of a hardware system that is generalize-able to more complex hardware systems is presented. This method is shown to reduce median tracking error on a soft robot hardware platform by 88%. Finally, particle swarm model predictive control (PSOMPC), a variant of NEMPC, is presented and modified to model and account for uncertainty in a dynamic system. Control performance of NEMPC and PSOMPC are presented for a set of control trials on simulated systems with uncertainty in parameters, states, and inputs, as well as on a soft robot hardware platform. PSOMPC is shown to have an increased robustness to system uncertainty, reducing expected collisions by 71% for a three-link robot arm with parameter uncertainty, input disturbances, and state measurement error.

Traditional robots, while capable of being efficient and effective for the task they were designed, are dangerous when operating in unmodeled environments or around humans. The field of soft robotics attempts to increase the safety of robots thus enabling them to operate in environments where traditional robots should not operate. Because of this, soft robots were developed with different goals in mind than traditional robots and as such the traditional metrics used to evaluate standard robots are not effective for evaluating soft robots. New metrics need to be developed for soft robots so that effective comparison and evaluations can be made. This dissertation attempts to lay the groundwork for that process through a survey on soft robot metrics. Additionally we propose six soft robot actuator metrics that can be used to evaluate and compare characteristics and performance of soft robot actuators. Data from eight different soft robot rotational actuators (five distinct designs) were used to evaluate these soft robot actuator metrics and show their utility. New models, control methods and estimation methods also need to be developed for soft robots. Many of the traditional methods and assumptions for modeling and controlling robotic systems are not able to provide the fidelity that is needed for soft robots to effectively complete useful tasks. This dissertation presents specific developments in each of these areas of soft robot metrics, modeling, control and estimation. We show several incremental improvements to soft robot dynamic models as well as how they were used in control methods for more precise control. We also demonstrate a method for linearizing high degree of freedom models so it can be simplified for use in faster control methods for better performance. Lastly, we present an improved continuum joint configuration estimation method that uses a linear combination of length measurements. All these developments combine to help build the "fundamental engineering framework" that is needed for soft robotics as well as helping to move robots out of their confined spaces and bring them into new unmodeled/unstructured environments.

In recent years, the field of Soft Robotics has grown exponentially resulting in a variety of different soft robot designs. A majority of the current soft robots can easily be split into two distinct categories: Navigation and Grasping. Navigation robots alter their body orientation to navigate around an environment. Grasping robots are designed to grasp a variety of unknown objects without damaging said object. However, only a few robots are able to demonstrate both aspects and even fewer robots are able to do both simultaneously. As thus, the goal of this thesis is to create a soft robot that is able to pick up and support an additional payload. This thesis will explore the challenges and difficulties that come with designing such a robot. For this thesis, we chose to simplify the manufacturing process making it easy to create and test different designs. We primarily used Pneumatic Network actuators for the majority of the soft robot. This allowed us to use a layered manufacturing approach to create the full robot. Finally, we split the robot into two main components which have their own purpose, which made it easy to test and design each component. Attached to this thesis are three different supplementary videos. The first one labeled "Walking Gaits" demonstrate how the robot is capable of moving forward. This video is comprised of several sections showing the full robot moving, just the base moving, and the full robot briefly moving as it supports a payload. The second video is labeled "Additional Walking". This video shows how the base can effectively move around a given environment. The final video if called "Grasping Method" which demonstrates the different grasping methods that the full robot uses to pick up objects. / 2021-06-03T00:00:00Z

Soft robotic platforms are becoming increasingly popular as they are generally safer, lighter, and easier to manufacture than their more rigid, heavy, traditional counterparts. These soft platforms, while inherently safer, come with significant drawbacks. Their compliant components are more difficult to model, and their underdamped nature makes them difficult to control. Additionally, they are so lightweight that a payload of just a few pounds has a significant impact on the manipulator dynamics. This thesis presents novel methods for addressing these issues. In previous research, Model Predictive Control has been demonstrably useful for joint angle control for these soft robots, using a rigid inverted pendulum model for each link. A model describing the dynamics of the entire arm would be more desirable, but with high Degrees of Freedom it is computationally expensive to optimize over such a complex model. This thesis presents a method for simplifying and linearizing the full-arm model (the Coupling-Torque method), and compares control performance when using this method of linearization against control performance for other linearization methods. The comparison shows the Coupling-Torque method yields good control performance for manipulators with seven or more Degrees of Freedom. The decoupled nature of the Coupling-Torque method also makes adaptive control, of the form described in this thesis, easier to implement. The Coupling-Torque method improves performance when the dynamics are known, but when a payload of unknown mass is attached to the end effector it has a significant impact on the dynamics. Adaptive Control is needed at this point to compensate for the model's poor approximation of the system. This thesis presents a method of layering Model Reference Adaptive Control in concert with Model Predictive Control that improves control performance in this scenario. The adaptive controller modifies dynamic parameters, which are then delivered to the optimizer, which then returns inputs for the system that take all of this information into account. This method has been shown to reduce step input tracking error by 50% when implemented on the soft robot.

This dissertation is focused on the modeling and robust model-based control of high degree-of-freedom (DoF) systems. While most of the contributions are applicable to any difficult-to-model system, this dissertation focuses specifically on applications to large-scale soft robots because their many joints and pressures constitute a high-DoF system and their inherit softness makes them difficult to model accurately. First a joint-angle estimation and kinematic calibration method for soft robots is developed which is shown to decrease the pose prediction error at the end of a 1.5 m robot arm by about 85\%. A novel dynamic modelling approach which can be evaluated within microseconds is then formulated for continuum type soft robots. We show that deep neural networks (DNNs) can be used to approximate soft robot dynamics given training examples from physics-based models like the ones described above. We demonstrate how these machine-learning-based models can be evaluated quickly to perform a form of optimal control called model predictive control (MPC). We describe a method of control trajectory parameterization that enables MPC to be applied to systems with more DoF and with longer prediction horizons than previously possible. We show that this parameterization decreases MPC's sensitivity to model error and drastically reduces MPC solve times. A novel form of MPC is developed based on an evolutionary optimization algorithm that allows the optimization to be parallelized on a computer's graphics processing unit (GPU). We show that this evolutionary MPC (EMPC) can greatly decrease MPC solve times for high DoF systems without large performance losses, especially given a large GPU. We combine the ideas of machine learned DNN models of robot dynamics, with parameterized and parallelized MPC to obtain a nonlinear version of EMPC which can be run at higher rates and find better solutions than many state-of-the-art optimal control methods. Finally we demonstrate an adaptive form of MPC that can compensate for model error or changes in the system to be controlled. This adaptive form of MPC is shown to inherit MPC's robustness to completely unmodeled disturbances and adaptive control's ability to decrease trajectory tracking errors over time.

Soft robots are inherently safer than traditional robots due to their compliance and high power density ratio resulting in lower accidental impact forces. Thus they are a natural option for human-robot interaction. This thesis specifically looked at human-robot co-manipulation which is defined as a human and a robot working together to move an object too large or awkward to be safely maneuvered by a single agent. To better understand how humans communicate while co-manipulating an object, this work looked at haptic interaction of human-human dyadic co-manipulation trials and studied some of the trends found in that interaction. These trends point to ways robots can effectively work with human partners in the future. Before successful human-robot co-manipulation with large-scale soft robots can be achieved, low-level joint angle control is needed. Low-level model predictive control of soft robot joints requires a sufficiently accurate model of the system. This thesis introduces a recursive Newton-Euler method for deriving the dynamics that is sufficiently accurate and accounts for flexible joints in an intuitive way. This model has been shown to be accurate to a median absolute error of 3.15 degrees for a three-link three-joint six degree of freedom soft robot arm. Once a sufficiently accurate model was developed, a gain-based evolutionary model predictive control (MPC) technique was formulated based on a previous evolutionary MPC technique. This new method is referred to as model evolutionary gain-based predictive control or MEGa-PC. This control law is compared to nonlinear evolutionary model predictive control (NEMPC). The new technique allows intentionally decreasing the control frequency to 10 Hz while maintaining control of the system. This is proven to help MPC solve more difficult problems by having the ability to extend the control horizon. This new controller is also demonstrated to work well on a three-joint three-link soft robot arm. Although complete physical human-robot co-manipulation is outside the scope of this thesis, this thesis covers three main building blocks for physical human and soft robot co-manipulation: human-human haptic communication, soft robot modeling, and model evolutionary gain-based predictive control.

Inflatable robots with pneumatic actuation are naturally lightweight and compliant. Both of these characteristics make a robot of this type better suited for human environments where unintentional impacts will occur. The dynamics of an inflatable robot are complex and dynamic models that explicitly allow variable stiffness control have not been well developed. In this thesis, a dynamic model was developed for an antagonistic, pneumatically actuated joint with inflatable links.The antagonistic nature of the joint allows for the control of two states, primarily joint position and stiffness. First a model was developed to describe the position states. The model was used with model predictive control (MPC) and linear quadratic control (LQR) to control a single degree of freedom platform to within 3° of a desired angle. Control was extended to multiple degrees of freedom for a pick and place task where the pick was successful ten out of ten times and the place was successful eight out of ten times.Based on a torque model for the joint which accounts for pressure states that was developed in collaboration with other members of the Robotics and Dynamics Lab at Brigham Young University, the model was extended to account for the joint stiffness. The model accounting for position, stiffness, and pressure states was fit to data collected from the actual joint and stiffness estimation was validated by stiffness measurements.Using the stiffness model, sliding mode control (SMC) and MPC methods were used to control both stiffness and position simultaneously. Using SMC, the joint stiffness was controlled to within 3 Nm/rad of a desired trajectory at steady state and the position was controlled to within 2° of a desired position trajectory at steady state. Using MPC,the joint stiffness was controlled to within 1 Nm/rad of a desired trajectory at steady state and the position was controlled to within 2° of a desired position trajectory at steady state. Stiffness control was extended to multiple degrees of freedom using MPC where each joint was treated as independent and uncoupled. Controlling stiffness reduced the end effecter deflection by 50% from an applied load when high stiffness (50 Nm/rad) was used rather than low stiffness (35 Nm/rad).This thesis gives a state space dynamic model for an inflatable, pneumatically actuated joint and shows that the model can be used for accurate and repeatable position and stiffness control with stiffness having a significant effect.

Soft lightweight robots provide an inherently safe solution to using robots in unmodeled environments by maintaining safety without increasing cost through expensive sensors. Unfortunately, many practical problems still need to be addressed before soft robots can become useful in real world tasks. Unlike traditional robots, soft robot geometry is not constant but can change with deflation and reinflation. Small errors in a robot's kinematic model can result in large errors in pose estimation of the end effector. This error, coupled with the inherent compliance of soft robots and the difficulty of soft robot joint angle sensing, makes it very challenging to accurately control the end effector of a soft robot in task space. However, this inherent compliance means that soft robots lend themselves nicely to coordinated multi-arm manipulation tasks, as deviations in end effector pose do not result in large force buildup in the arms or in the object being manipulated. Coordinated, multi-arm manipulation with soft robots is the focus of this thesis. We first developed two tools enabling multi-arm manipulation with soft robots: (1) a hybrid servoing control scheme for task space control of soft robot arms, and (2) a general base placement optimization for the robot arms in a multi-arm manipulation task. Using these tools, we then developed and implemented a simple multi-arm control scheme. The hybrid servoing control scheme combines inverse kinematics, joint angle control, and task space servoing in order to reduce end effector pose error. We implemented this control scheme on two soft robots and demonstrated its effectiveness in task space control. Having developed a task space controller for soft robots, we then approached the problem of multi-arm manipulation. The placement of each arm for a multi-arm task is non-trivial. We developed an evolutionary optimization that finds the optimal arm base location for any number of user-defined arms in a user-defined task or workspace. We demonstrated the utility of this optimization in simulation, and then used it to determine the arm base locations for two arms in two real world coordinated multi-arm manipulation tasks. Finally, we developed a simple multi-arm control scheme for soft robots and demonstrated its effectiveness using one soft robot arm, and one rigid robot with low-impedance torque control. We placed each arm base in the pose determined by the base placement optimization, and then used the hybrid servoing controller in our multi-arm control scheme to manipulate an object through two desired trajectories.

This paper deals with a novel active bending soft body that employs metamaterials and combines soft behavior, integrated actuation, low complexity and a high density of producible forces and moments. The presented concept consists of a tube-like structure with tailored, unconventional material properties which enable the generation of a bending deformation and/or moment when circumferential stress and/or strain is induced. Circumferential actuation can be generated by a difference in pressure between the internal and external surface of the tube or, alternatively, by distributed expansion actuators that act radially or tangentially (e.g. shape memory wires). In addition to an analytical model, this paper also presents a design procedure and deals with the implementation of the proposed concept in a functional prototype and its experimental characterization.