The purpose of this dissertation is to develop a design technique for fractional PID controllers to achieve a closed loop sensitivity bandwidth approximately equal to a desired bandwidth using frequency loop shaping techniques. This dissertation analyzes the effect of the order of a fractional integrator which is used as a target on loop shaping, on stability and performance robustness. A comparison between classical PID controllers and fractional PID controllers is presented. Case studies where fractional PID controllers have an advantage over classical PID controllers are discussed. A frequency-domain loop shaping algorithm is developed, extending past results from classical PID’s that have been successful in tuning controllers for a variety of practical systems.

The interaction between humans and robots has become an important area of research as the diversity of robotic applications has grown. The cooperation of a human and robot to achieve a goal is an important area within the physical human-robot interaction (pHRI) field. The expansion of this field is toward moving robotics into applications in unstructured environments. When humans cooperate with each other, often there are leader and follower roles. These roles may change during the task. This creates a need for the robotic system to be able to exchange roles with the human during a cooperative task. The unstructured nature of the new applications in the field creates a need for robotic systems to be able to interact in six degrees of freedom (DOF). Moreover, in these unstructured environments, the robotic system will have incomplete information. This means that it will sometimes perform an incorrect action and control methods need to be able to correct for this. However, the most compelling applications for robotics are where they have capabilities that the human does not, which also creates the need for robotic systems to be able to correct human action when it detects an error. Activity in the brain precedes human action. Utilizing this activity in the brain can classify the type of interaction desired by the human. For this dissertation, the cooperation between humans and robots is improved in two main areas. First, the ability for electroencephalogram (EEG) to determine the desired cooperation role with a human is demonstrated with a correct classification rate of 65%. Second, a robotic controller is developed to allow the human and robot to cooperate in six DOF with asymmetric role exchange. This system allowed human-robot cooperation to perform a cooperative task at 100% correct rate. High, medium, and low levels of robotic automation are shown to affect performance, with the human making the greatest numbers of errors when the robotic system has a medium level of automation.

This study presents quantification of ankle stability as affected by environmental conditions in two degrees of freedom (DOF) with three distinct analysis techniques. Additionally, this study presents gender-specific trends for comparison. Intuitively, ankle stability decreased in less stable environments with a negative simulated stiffness. Female subjects generally suffered a greater loss of stability in moderately and highly unstable environments. Both gender groups exhibited greater stability in the sagittal plane than the frontal plane across the entire range of simulated stiffness's. Outcomes of this study are useful in the design of controllers for lower extremity physically-interactive robotics, understanding situations in which the ankle is likely to lose stability, and understanding the strengths and weaknesses of unique analysis techniques.

The aim of this project was to develop user-friendly methods for programming and controlling a new type of small robot platform, called Pheeno, both individually and as part of a group. Two literature reviews are presented to justify the need for these robots and to discuss what other platforms have been developed for similar applications. In order to accomplish control of multiple robots work was done on controlling a single robot first. The response of a gripper arm attachment for the robot was smoothed, graphical user interfaces were developed, and commands were sent to a single robot using a video game controller. For command of multiple robots a class was developed in Python to make it simpler to send commands and keep track of different characteristics of each individual robot. A simple script was also created as a proof of concept to show how threading could be used to send different commands simultaneously to multiple robots in order to test algorithms on a group of robots. The class and two other scripts necessary for implementing the class are also presented to make it possible for future use of the given work.

Research on human grasp typically involves the grasp of objects designed for the study of fingertip forces. Instrumented objects for such studies have often been designed for the simulation of functional tasks, such as feeding oneself, or for rigidity such that the objects do not deform when grasped. The goal of this thesis was to design a collapsible, instrumented object to study grasp of breakable objects. Such an object would enable experiments on human grip responses to unexpected finger-object events as well as anticipatory mechanisms once object fragility has been observed. The collapsible object was designed to be modular to allow for properties such as friction and breaking force to be altered. The instrumented object could be used to study both human and artificial grasp.

The quality of life of many people is lowered by impediments to walking ability caused by neurological conditions such as strokes. Since the ankle joint plays an important role in locomotion, it is a common subject of study in rehabilitation research. Robotic devices such as active ankle-foot orthoses and powered exoskeletons have the potential to be used directly in physical therapy or indirectly in research pursuing more effective rehabilitation methods. This paper presents the LiTREAD, a lightweight three degree-of-freedom robotic exoskeletal ankle device. This novel robotic system is designed to be worn on a user's leg and actuate the foot position during treadmill studies. The robot's sagittal plane actuation is complemented by passive virtual axis systems in the frontal and transverse planes. Together, these degrees of freedom allow the device to approximate the full range of motion of the ankle. The virtual axis mechanisms feature locking configurations that will allow the effect of these degrees of freedom on gait dynamics to be studied. Based on a kinematic analysis of the robot's actuation and geometry, it is expected to meet and exceed its torque and speed targets, respectively. The device will fit either leg of a range of subject sizes, and is expected to weigh just 1.3 kg (2.9 lb.). These features and characteristics are designed to minimize the robot's interference with the natural walking motion. Pending validation studies confirming that all design criteria have been met, the LiTREAD prototype that has been constructed will be utilized in various experiments investigating properties of the ankle such as its mechanical impedance. It is hoped that the LiTREAD will yield valuable data that will expand our knowledge of the ankle and aid in the design of future lower-extremity devices.

For the past two decades, advanced Limb Gait Simulators and Exoskeletons have been developed to improve walking rehabilitation. A Limb Gait Simulator is used to analyze the human step cycle and/or assist a user walking on a treadmill. Most modern limb gait simulators, such as ALEX, have proven themselves effective and reliable through their usage of motors, springs, cables, elastics, pneumatics and reaction loads. These mechanisms apply internal forces and reaction loads to the body. On the other hand, external forces are those caused by an external agent outside the system such as air, water, or magnets. A design for an exoskeleton using external forces has seldom been attempted by researchers. This thesis project focuses on the development of a Limb Gait Simulator based on a Pure External Force and has proven its effectiveness in generating torque on the human leg. The external force is generated through air propulsion using an Electric Ducted Fan (EDF) motor. Such a motor is typically used for remote control airplanes, but their applications can go beyond this. The objective of this research is to generate torque on the human leg through the control of the EDF engines thrust and the opening/closing of the reverse thruster flaps. This device qualifies as "assist as needed"; the user is entirely in control of how much assistance he or she may want. Static thrust values for the EDF engine are recorded using a thrust test stand. The product of the thrust (N) and the distance on the thigh (m) is the resulting torque. With the motor running at maximum RPM, the highest torque value reached was that of 3.93 (Nm). The motor EDF motor is powered by a 6S 5000 mAh LiPo battery. This torque value could be increased with the usage of a second battery connected in series, but this comes at a price. The designed limb gait simulator demonstrates that external forces, such as air, could have potential in the development of future rehabilitation devices.

Walking ability is a complex process that is essential to humans, critical for performing a range of everyday tasks and enables a healthy, independent lifestyle. Human gait has evolved to be robust, adapting to a wide range of external stimuli, including variable walking surface compliance. Unfortunately, many people suffer from impaired gait as a result of conditions such as stroke. For these individuals, recovering their gait is a priority and a challenge. The ASU Variable Stiffness Treadmill (VST) is a device that is able to the change its surface compliance through its unique variable stiffness mechanism. By doing this, the VST can be used to investigate gait and has potential as a rehabilitation tool. The objective of this research is to design a variable damping mechanism for the VST, which addresses the need to control effective surface damping, the only form of mechanical impedance that the VST does not currently control. Thus, this project will contribute toward the development of the Variable Impedance Treadmill (VIT), which will encompass a wider range of variable surface compliance and enable all forms of impedance to be con- trolled for the first time. To achieve this, the final design of the mechanism will employ eddy current damping using several permanent magnets mounted to the treadmill and a large copper plate stationed on the ground. Variable damping is obtained by using lead screw mechanisms to remove magnets from acting on the copper plate, which effectively eliminates their effect on damping and changes the overall treadmill surface damping. Results from experimentation validate the mechanism's ability to provide variable damping to the VST. A model for effective surface damping is generated based on open-loop characterization experiments and is generalized for future experimental setups. Overall, this project progresses to the development of the VIT and has potential applications in walking surface simulation, gait investigation, and robot-assisted rehabilitation technology.

In recent years, environment mapping has garnered significant interest in both industrial and academic settings as a viable means of generating comprehensive virtual models of the physical world. These maps are created using simultaneous localization and mapping (SLAM) algorithms that combine depth contours with visual imaging information to create rich, layered point clouds. Given the recent advances in virtual reality technology, these generated point clouds can be imported onto the Oculus Rift or similar headset for virtual reality implementation. This project deals with the robotic implementation of RGB-D SLAM algorithms on mobile ground robots to generate complete point clouds that can be processed off-line and imported into virtual reality engines for viewing in the Oculus Rift. This project uses a ground robot along with a Kinect sensor to collect RGB-D data of the surrounding environment to build point cloud maps using SLAM software. These point clouds are then exported as object or polygon files for post-processing in software engines such as Meshlab or Unity. The point clouds generated from the SLAM software can be viewed in the Oculus Rift as is. However, these maps are mainly empty space and can be further optimized for virtual viewing. Additional techniques such as meshing and texture meshing were implemented on the raw point cloud maps and tested on the Oculus Rift. The aim of this project was to increase the potential applications for virtual reality by taking a robotic mapping approach to virtual reality environment development. This project was successful in achieving its objective. The following report details the processes used in developing a remotely-controlled robotic platform that can scan its environment and generate viable point cloud maps. These maps are then processed off line and ported into virtual reality software for viewing through the Oculus Rift.

As robots become more prevalent, the need is growing for efficient yet stable control systems for applications with humans in the loop. As such, it is a challenge for scientists and engineers to develop robust and agile systems that are capable of detecting instability in teleoperated systems. Despite how much research has been done to characterize the spatiotemporal parameters of human arm motions for reaching and gasping, not much has been done to characterize the behavior of human arm motion in response to control errors in a system. The scope of this investigation is to investigate human corrective actions in response to error in an anthropomorphic teleoperated robot limb. Characterizing human corrective actions contributes to the development of control strategies that are capable of mitigating potential instabilities inherent in human-machine control interfaces. Characterization of human corrective actions requires the simulation of a teleoperated anthropomorphic armature and the comparison of a human subject's arm kinematics, in response to error, against the human arm kinematics without error. This was achieved using OpenGL software to simulate a teleoperated robot arm and an NDI motion tracking system to acquire the subject's arm position and orientation. Error was intermittently and programmatically introduced to the virtual robot's joints as the subject attempted to reach for several targets located around the arm. The comparison of error free human arm kinematics to error prone human arm kinematics revealed an addition of a bell shaped velocity peak into the human subject's tangential velocity profile. The size, extent, and location of the additional velocity peak depended on target location and join angle error. Some joint angle and target location combinations do not produce an additional peak but simply maintain the end effector velocity at a low value until the target is reached. Additional joint angle error parameters and degrees of freedom are needed to continue this investigation.