In this thesis, the problem of designing model discrimination algorithms for unknown nonlinear systems is considered, where only raw experimental data of the system is available. This kind of model discrimination techniques finds one of its application in the estimation of the system or intent models under consideration, where all incompatible models are invalidated using new data that is available at run time. The proposed steps to reach the end goal of the algorithm for intention estimation involves two steps: First, using available experimental data of system trajectories, optimization-based techniques are used to over-approximate/abstract the dynamics of the system by constructing an upper and lower function which encapsulates/frames the true unknown system dynamics. This over-approximation is a conservative preservation of the dynamics of the system, in a way that ensures that any model which is invalidated against this approximation is guaranteed to be invalidated with the actual model of the system. The next step involves the use of optimization-based techniques to investigate the distinguishability of pairs of abstraction/approximated models using an algorithm for 'T-Distinguishability', which gives a finite horizon time 'T', within which the pair of models are guaranteed to be distinguished, and to eliminate incompatible models at run time using a 'Model Invalidation' algorithm. Furthermore, due the large amount of data under consideration, some computation-aware improvements were proposed for the processing of the raw data and the abstraction and distinguishability algorithms.The effectiveness of the above-mentioned algorithms is demonstrated using two examples. The first uses the data collected from the artificial simulation of a swarm of agents, also known as 'Boids', that move in certain patterns/formations, while the second example uses the 'HighD' dataset of naturalistic trajectories recorded on German Highways for vehicle intention estimation.

In the development of autonomous ground vehicles (AGVs), how to guarantee vehicle lateral stability is one of the most critical aspects. Based on nonlinear vehicle lateral and tire dynamics, new driving requirements of AGVs demand further studies and analyses of vehicle lateral stability control strategies. To achieve comprehensive analyses and stability-guaranteed vehicle lateral driving control, this dissertation presents three main contributions.First, a new method is proposed to estimate and analyze vehicle lateral driving stability regions, which provide a direct and intuitive demonstration for stability control of AGVs. Based on a four-wheel vehicle model and a nonlinear 2D analytical LuGre tire model, a local linearization method is applied to estimate vehicle lateral driving stability regions by analyzing vehicle local stability at each operation point on a phase plane. The obtained stability regions are conservative because both vehicle and tire stability are simultaneously considered. Such a conservative feature is specifically important for characterizing the stability properties of AGVs. Second, to analyze vehicle stability, two novel features of the estimated vehicle lateral driving stability regions are studied. First, a shifting vector is formulated to explicitly describe the shifting feature of the lateral stability regions with respect to the vehicle steering angles. Second, dynamic margins of the stability regions are formulated and applied to avoid the penetration of vehicle state trajectory with respect to the region boundaries. With these two features, the shiftable stability regions are feasible for real-time stability analysis. Third, to keep the vehicle states (lateral velocity and yaw rate) always stay in the shiftable stability regions, different control methods are developed and evaluated. Based on different vehicle control configurations, two dynamic sliding mode controllers (SMC) are designed. To better control vehicle stability without suffering chattering issues in SMC, a non-overshooting model predictive control is proposed and applied. To further save computational burden for real-time implementation, time-varying control-dependent invariant sets and time-varying control-dependent barrier functions are proposed and adopted in a stability-guaranteed vehicle control problem. Finally, to validate the correctness and effectiveness of the proposed theories, definitions, and control methods, illustrative simulations and experimental results are presented and discussed.

This paper introduces a variable impedance controller which dynamically modulates both its damping and stiffness to improve the trade-off between stability and agility in coupled human-robot systems and reduce the human user’s effort. The controller applies a range of robotic damping from negative to positive values to either inject or dissipate energy based on the user’s intent of motion. The controller also estimates the user’s intent of direction and applies a variable stiffness torque to stabilize the user towards an estimated ideal trajectory. To evaluate the controller’s ability to improve the stability/agility trade-off and reduce human effort, a study was designed for human subjects to perform a 2D target reaching task while coupled with a wearable ankle robot. A constant impedance condition was selected as a control with which to compare the variable impedance condition. The position, speed, and muscle activation responses were used to quantify the user’s stability, agility, and effort, respectively. Stability was quantified spatially and temporally, with both overshoot and stabilization time showing no statistically significant difference between the two experimental conditions. Agility was quantified using mean and maximum speed, with both increasing from the constant impedance to variable impedance condition by 29.8% and 59.9%, respectively. Effort was quantified by the overall and maximum muscle activation data, both of which showed a ~10% reduction in effort. Overall, the study demonstrated the effectiveness of the variable impedance controller.

Bicycle stabilization has become a popular topic because of its complex dynamic behavior and the large body of bicycle modeling research. Riding a bicycle requires accurately performing several tasks, such as balancing and navigation which may be difficult for disabled people. Their problems could be partially reduced by providing steering assistance. For stabilization of these highly maneuverable and efficient machines, many control techniques have been applied – achieving interesting results, but with some limitations which includes strict environmental requirements. This thesis expands on the work of Randlov and Alstrom, using reinforcement learning for bicycle self-stabilization with robotic steering. This thesis applies the deep deterministic policy gradient algorithm, which can handle continuous action spaces which is not possible for Q-learning technique. The research involved algorithm training on virtual environments followed by simulations to assess its results. Furthermore, hardware testing was also conducted on Arizona State University’s RISE lab Smart bicycle platform for testing its self-balancing performance. Detailed analysis of the bicycle trial runs are presented. Validation of testing was done by plotting the real-time states and actions collected during the outdoor testing which included the roll angle of bicycle. Further improvements in regard to model training and hardware testing are also presented.

The human shoulder plays an integral role in upper limb motor function. As the basis of arm motion, its performance is vital to the accomplishment of daily tasks. Impaired motor control, as a result of stroke or other disease, can cause errors in shoulder position to accumulate and propagate to the entire arm. This is why it is a highlight of concern for clinicians and why it is an important point of study. One of the primary causes of impaired shoulder motor control is abnormal mechanical joint impedance, which can be modeled as a 2nd order system consisting of mass, spring and damper. Quantifying shoulder stiffness and damping between healthy and impaired subjects could help improve our collective understanding of how many different neuromuscular diseases impact arm performance. This improved understanding could even lead to better rehabilitation protocols for conditions such as stroke through better identification and targeting of damping dependent spasticity and stiffness dependent hypertonicity. Despite its importance, there is a fundamental knowledge gap in the understanding of shoulder impedance, mainly due to a lack of appropriate characterization tools. Therefore, in order to better quantify shoulder stiffness and damping, a novel low-inertia shoulder exoskeleton is introduced in this work. The device was developed using a newly pioneered parallel actuated robot architecture specifically designed to interface with complex biological joints like the shoulder, hip, wrist and ankle. In addition to presenting the kinematics and dynamics of the shoulder exoskeleton, a series of validation experiments are performed on a human shoulder mock-up to quantify its ability to estimate known impedance properties. Finally, some preliminary data from human experiments is provided to demonstrate the device’s ability to collect the measurements needed to estimate shoulder stiffness and damping while worn by a subject.

Control algorithm development for quadrotor is usually based solely on rigid body dynamics neglecting aerodynamics. Recent work has demonstrated that such a model is suited only when operating at or near hover conditions and low-speed flight. When operating in confined spaces or during aggressive maneuvers destabilizing forces and moments are induced due to aerodynamic effects. Studies indicate that blade flapping, induced drag, and propeller drag influence forward flight performance while other effects like vortex ring state, ground effect affect vertical flight performance. In this thesis, an offboard data-driven approach is used to derive models for parasitic (bare-airframe) drag and propeller drag. Moreover, thrust and torque coefficients are identified from static bench tests. Among the two, parasitic drag is compensated for in the position controller module in the PX4 firmware. 2-D circular, straight line, and minimum snap rectangular trajectories with corridor constraints are tested exploiting differential flatness property wherein altitude and yaw angle are constant. Flight tests are conducted at ASU Drone Studio and results of tracking performance with default controller and with drag compensated position controller are presented. Root mean squared tracking error in individual axes is used as a metric to evaluate the model performance. Results indicate that, for circular trajectory, the root mean squared error in the x-axis has reduced by 44.54% and in the y-axis by 39.47%. Compensation in turn degrades the tracking in both axis by a maximum under 12% when compared to the default controller for rectangular trajectory case. The x-axis tracking error for the straight-line case has improved by 44.96% with almost no observable change in the y-axis.

The objective of this project was to research and experimentally test methods of localization, waypoint following, and actuation for high-speed driving by an autonomous vehicle. This thesis describes the implementation of LiDAR localization techniques, Model Predictive Control waypoint following, and communication for actuation on a 2016 Chevrolet Camaro, Arizona State University’s former EcoCAR. The LiDAR localization techniques include the NDT Mapping and Matching algorithms from the open-source autonomous vehicle platform, Autoware. The mapping algorithm was supplemented by that of Google Cartographer due to the limitations of map size in Autoware’s algorithms. The Model Predictive Control for waypoint following and the computer-microcontroller-actuator communication line are described. In addition to this experimental work, the thesis discusses an investigation of alternative approaches for each problem.

This paper presents a variable damping controller that can be implemented into wearable and exoskeleton robots. The variable damping controller functions by providing different levels of robotic damping from negative to positive to the coupled human-robot system. The wearable ankle robot was used to test this control strategy in the different directions of motion. The range of damping applied was selected based on the known inherent damping of the human ankle, ensuring that the coupled system became positively damped, and therefore stable. Human experiments were performed to understand and quantify the effects of the variable damping controller on the human user. Within the study, the human subjects performed a target reaching exercise while the ankle robot provided the system with constant positive, constant negative, or variable damping. These three damping conditions could then be compared to analyze the performance of the system. The following performance measures were selected: maximum speed to quantify agility, maximum overshoot to quantify stability, and muscle activation to quantify effort required by the human user. Maximum speed was found to be statistically the same in the variable damping controller and the negative damping condition and to be increased from positive damping controller to variable damping condition by 57.9%, demonstrating the agility of the system. Maximum overshoot was found to significantly decrease overshoot from the negative damping condition to the variable damping controller by 39.6%, demonstrating an improvement in system stability with the variable damping controller. Muscle activation results showed that the variable damping controller required less effort than the positive damping condition, evidenced by the decreased muscle activation of 23.8%. Overall, the study demonstrated that a variable damping controller can balance the trade-off between agility and stability in human-robot interactions and therefore has many practical implications.

Autonomous vehicle technology has been evolving for years since the Automated Highway System Project. However, this technology has been under increased scrutiny ever since an autonomous vehicle killed Elaine Herzberg, who was crossing the street in Tempe, Arizona in March 2018. Recent tests of autonomous vehicles on public roads have faced opposition from nearby residents. Before these vehicles are widely deployed, it is imperative that the general public trusts them. For this, the vehicles must be able to identify objects in their surroundings and demonstrate the ability to follow traffic rules while making decisions with human-like moral integrity when confronted with an ethical dilemma, such as an unavoidable crash that will injure either a pedestrian or the passenger.

Testing autonomous vehicles in real-world scenarios would pose a threat to people and property alike. A safe alternative is to simulate these scenarios and test to ensure that the resulting programs can work in real-world scenarios. Moreover, in order to detect a moral dilemma situation quickly, the vehicle should be able to identify objects in real-time while driving. Toward this end, this thesis investigates the use of cross-platform training for neural networks that perform visual identification of common objects in driving scenarios. Here, the object detection algorithm Faster R-CNN is used. The hypothesis is that it is possible to train a neural network model to detect objects from two different domains, simulated or physical, using transfer learning. As a proof of concept, an object detection model is trained on image datasets extracted from CARLA, a virtual driving environment, via transfer learning. After bringing the total loss factor to 0.4, the model is evaluated with an IoU metric. It is determined that the model has a precision of 100% and 75% for vehicles and traffic lights respectively. The recall is found to be 84.62% and 75% for the same. It is also shown that this model can detect the same classes of objects from other virtual environments and real-world images. Further modifications to the algorithm that may be required to improve performance are discussed as future work.

Admittance control with fixed damping has been a successful control strategy in previous human-robotic interaction research. This research implements a variable damping admittance controller in a 7-DOF robotic arm coupled with a human subject’s arm at the end effector to study the trade-off of agility and stability and aims to produce a control scheme which displays both fast rise time and stability. The variable damping controller uses a measure of intent of movement to vary damping to aid the user’s movement to a target. The range of damping values is bounded by incorporating knowledge of a human arm to ensure the stability of the coupled human-robot system. Human subjects completed experiments with fixed positive, fixed negative, and variable damping controllers to evaluate the variable damping controller’s ability to increase agility and stability. Comparisons of the two fixed damping controllers showed as fixed damping increased, the coupled human-robot system reacted with less overshoot at the expense of rise time, which is used as a measure of agility. The inverse was also true; as damping became increasingly negative, the overshoot and stability of the system was compromised, while the rise time became faster. Analysis of the variable damping controller demonstrated humans could extract the benefits of the variable damping controller in its ability to increase agility in comparison to a positive damping controller and increase stability in comparison to a negative damping controller.