Mechanical Engineering
Permanent URI for this communityhttp://hdl.handle.net/1903/2263
Browse
43 results
Search Results
Item Metareasoning Strategies to Correct Navigation Failures of Autonomous Ground Robots(2024) Molnar, Sidney Leigh; Herrmann, Jeffrey; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Due to the complexity of autonomous systems, theoretically perfect path planning algorithms sometimes fail due to emergent behaviors that arise when interacting with different perception, mapping and goal planning subprocesses. These failures prevent mission success, especially in complex environments that have not previously been explored by the robot. To overcome these failures, many researchers have sought to develop parameter learning methods to improve either mission success or path planning convergence. Metareasoning, which can be simply described as “thinking about thinking,” offers another possible solution for mitigating these planning failures. This project offers a novel metareasoning approach that uses different methods of monitoring and control to detect and overcome path planning irregularities that contribute to path planning failures. All methods for the approaches were implemented as a part of the ARL ground autonomy stack which uses both global and local path planning ROS nodes. The proposed monitoring methods include listening to messages published to the system by the planning algorithms themselves, evaluating for the environmental context that the robot is in, the expected progress methods which use the robot’s movement capabilities to evaluate for progress that has been made from a milestone checkpoint, and the fixed radius methods which use user-selected parameters based on mission objectives to evaluate for the progress that has been made from a milestone checkpoint. The proposed control policies are the metric-based sequential policies which use benchmark robot performance metrics to select the order in which the planner combinations are to be launched, the context-based pairs policies which evaluate what happens when switching between only two planner combinations, and the restart policy which simply relaunches a new instance of the same planner combination. The study evaluated which monitoring and control policies, when paired, contributed to improved navigation performance and which policies contributed to degraded navigation performance by evaluating how close the robot was able to get to the final mission goal. Although specific methods were evaluated, the contributions of the project extend beyond the results by offering both a template for metareasoning approaches with regard to navigation as well as replicable algorithms that may be applied to any autonomous ground robot system. Additionally, this thesis presents ideas for additional research in order to determine under which conditions metareasoning will improve navigation.Item ON DATA-BASED MAPPING AND NAVIGATION OF UNMANNED GROUND VEHICLES(2024) Herr, Gurtajbir Singh; Chopra, Nikhil; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Unmanned ground vehicles (UGVs) have seen tremendous advancement in their capabilities and applications in the past two decades. With several key algorithmic and hardware breakthroughs and advancements in deep learning, UGVs are quickly becoming ubiquitous (finding applications as self-driving cars, for remote site inspections, in hospitals and shopping malls, among several others). Motivated by their large-scale adoption, this dissertation aims to enable the navigation of UGVs in complex environments. In this dissertation, a supervised learning-based navigation algorithm that utilizes model predictive control (MPC) for providing training data is developed. Improving MPC performance by data-based modelling of complex vehicle dynamics is then addressed. Finally, this dissertation deals with detecting and registering transparent objects that may deteriorate navigation performance. Navigation in dynamic environments poses unique challenges, particularly due to the limited knowledge of the decisions made by other agents and their objectives. In this dissertation, a solution that utilizes an MPC-based planner as an \textit{expert} to generate high-quality motion commands for a car-like robot operating in a simulated dynamic environment is proposed. These commands are then used to train a deep neural network, which learns to navigate. The deep learning-based planner is further enhanced with safety margins to improve its effectiveness in collision avoidance. The performance of the proposed method through simulations and real-world experiments, demonstrating its superiority in terms of obstacle avoidance and successful mission completion is showcased. This research has practical implications for the development of safer and more efficient autonomous vehicles. Many real-world applications rely on MPC to control UGVs due to its safety guarantees and constraint satisfaction properties. However, the performance of such MPC-based solutions is heavily reliant on the accuracy of the motion model. This dissertation addresses this challenge by exploring a data-based approach to discovering vehicle dynamics. Unlike existing physics-based models that require extensive testing setups and manual tuning for new platforms and driving surfaces, our approach leverages the universal differential equations (UDEs) framework to identify unknown dynamics from vehicle data. This innovative approach, which does not make assumptions about the unknown dynamics terms and directly models the vector field, is then deployed to showcase its efficacy. This research opens up new possibilities for more accurate and adaptable motion models for UGVs. With the increasing adoption of glass and other transparent materials, UGVS must be able to detect and register them for reliable navigation. Unfortunately, such objects are not easily detected by LiDARs and cameras. In this dissertation, algorithms for detecting and including glass objects in a Graph SLAM framework were studied. A simple and computationally inexpensive glass detection scheme to detect glass objects is utilized. The methodology to incorporate the identified objects into the occupancy grid maintained by such a framework is the presented. The issue of \textit{drift accumulation} that can affect mapping performance when operating in large environments is also addressed.Item Dynamic Control of Dexterous Soft Robotic Systems(2023) Weerakoon, Weerakoon Mudiyanselage Lasitha Tharinda; Chopra, Nikhil; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Soft robotics has grown exponentially during the past two decades due to the possibility of expanded manipulation capabilities over existing rigid robots in complex, unstructured environments. Additionally, soft robots can mitigate current safety risks associated with rigid robots due to their softness. The inspiration for soft robotics has been mainly due to the many examples from nature, such as the agile environmental interactions of the elephant trunk and octopus tentacles. Over the past two decades, several applications ranging from underwater operations to minimally invasive surgeries to space operations have been identified for soft robots. Motivated by these, the overall objective of this dissertation is to study and develop control frameworks for high-fidelity motion control of soft robotic systems. This entails exploiting generalized dynamics models for robust/adaptive control strategies for achieving various operational tasks involved in non-ideal environments, utilizing integrated sensing technologies, and investigating control of underactuated soft robotic systems. This dissertation delve into passivity-based adaptive task space control for soft robots, mitigating uncertainty in the parameters as accurate parameter estimation is particularly hard in soft robotic systems. Further, this approach is extended to task space bilateral teleoperation of a soft follower-rigid leader system exploiting null space velocity tracking to achieve sub-task goals such as conforming to the degree of curvature limits in the soft robot. An enhanced dynamics model is also introduced tailored for planar soft robots and elaborate on passivity-based robust control methods for task space trajectory tracking within this context. This enhanced dynamics model is subsequently extended to encompass 3D spatial soft robots and a comprehensive framework for passivity-based robust task space bilateral teleoperation is discussed. Extensive numerical simulations and experiments are conducted to illustrate the efficacy of these proposed control frameworks. Moreover, to deploy soft robots in the real world, this dissertation study integrated sensing and control of soft robots and a stretchable soft-sensing skin for proprioception s introduced. The mapping from the strain signal to the curvature degree is estimated using a recurrent neural network. Further, an adaptive control framework for curvature tracking is proposed, leveraging the soft stretchable sensing skins and providing experimental evidence of its successful application. This dissertation also introduces a novel robotic system known as the hybrid rigid-soft robot, composed of serially attached rigid and soft links, offering a fusion of the dexterity inherent to soft robots with the precision and payload capacity associated with rigid counterparts. Notably, the study demonstrates that well-established passivity-based adaptive and robust control techniques can effectively apply to this unique class of robots. A soft inverted pendulum with a revolute base is also introduced, establishing a scientific foundation and a methodological approach for introducing innovative soft robots in various practical applications. An energy-based controller is discussed for the swing-up and stabilization of the soft inverted pendulum system, highlighting the efficacy of the controller through simulations. Further, a comprehensive control architecture is developed for the swing-up and stabilization of a class of underactuated mechanical systems, including the soft inverted pendulum, by applying output partial feedback linearization and linear control techniques that avoid switching between controllers. The utility of this control architecture is illustrated using numerical simulations on the soft inverted pendulum. These research endeavors collectively contribute to advancing the understanding of soft robotics and developing effective control strategies for various dexterous soft robotic systems.Item MULTI-AGENT SPATIAL COORDINATION VIA TIME-VARIATIONS IN COVERAGE CONTROL(2023) Xu, Xiaotian; Diaz-Mercado, Yancy; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Coverage control of multi-agent systems (MASs) spatially spreads out a group of agents to form a configuration over a domain of interest. This research investigates the two fundamental elements embedded in coverage control, i.e., the time-varying density function and the time-varying domain, and how these can be leveraged to achieve collaborative controls of MASs. We focus on three problems: first, we abstract a robotic swarm, so it can be controlled as a whole where the robotic team adaptively finds the suitable spatial configuration; second, such abstraction of a MAS is extended to a higher-dimensional embedding for an interactive multi-agent aerial cinematography application; and third, a multi-objective formulation is developed to spatially distribute a MAS and take advantage of its collective effort to persistently cover a space. In contrast to the coverage with time-varying densities which has been actively studied, we address the coverage control over time-varying domains in the first problem, so the control of a MAS, in terms of its position, scale, shape, etc., is enabled and is simplified into manipulating the domain to be covered directly. The agents coordinate themselves to accommodate the evolution of the domain, even when the domain is evolving fast. A MAS control algorithm, named Swarm Herding, which is built upon the proposed control mechanism is implemented. In pursuit of this approach, contributions are made to the problem of coverage control over time-varying convex and non-convex domains for abstracting the swarm, efficiently tracking the evolution of the domains, and synthesizing the specialized controllers for every agent in the swarm. In the second, the abstraction is extended to a hemispherical manifold under the geodesic metric, and it is employed to enable an interactive motion coordinator for multi-robot aerial cinematography. The emphases are on collaborative behavior for multiple unmanned aerial vehicles (UAVs), dynamic target tracking, and real-time interaction for aesthetic cinematography objectives. Contributions are made in accommodating the realistic issues that occurred in MAS coverage over manifolds embedded in a higher dimension and in the design of a distributed interactive framework to provide high-level position instructions for a group of UAVs which addresses the gap in the ``one-pilot-many-robot'' feature. In the third problem, a multi-objective coverage control of MASs is formulated to take advantage of the collective effort of a team of mobile sensors to persistently explore a domain of interest. In addition to the standard locational coverage objective, a new perceptional coverage objective is introduced to drive agents around in the domain to gain information. The collaboration between agents is defined not only in terms of exchanging the knowledge of the domain as in previous work but also in terms of inter-agent motion coordination which reduces redundant visits to certain locations by agents. Contributions are made with respect to information exchange with performance guarantees, multi-objective coverage control of MASs with time-varying state-dependent density functions, and analysis of the effects of the multiple objectives.Item Modeling and Characterization of Bioinspired Hybrid Flapping/Gliding Flight for Flapping Wing Air Vehicles(2022) Johnson, Lena; Bruck, Hugh; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Unmanned Aerial Vehicles (UAVs) are increasingly being used for applications that require longer, reliable flight duration and distances. The greatest limitation to achieving these desired flights is the current on board battery technology which, restricted by internal chemistry and external size, can only provide a finite amount of power over time. Efforts to increase the battery’s efficiency and energy storage tend to rely on cumbersome methods that add weight and/or complexity to the system. However natural flyers, though also limited by a finite amount of internal energy gained through food consumption, are able to extend their flights through techniques that either utilize their inherent aerodynamic advantages or advantageously employ atmospheric phenomena. Flapping-Wing UAVs (FWUAVs) are as limited by their onboard battery as any other type of UAV, but because of their bio-inspired functionality are uniquely suited to utilize natural flight extension methods. Therefore, this PhD presents an analysis of the exploration of bio-inspired, hybrid flapping/gliding, also known as intermittent gliding, techniques to improve the flight performance of a FWUAV. Robo Raven is the FWUAV that was chosen as the research platform for this work. It was developed by researchers at the University of Maryland to perform prolonged, untethered flights and exhibit a flight proficiency that combined the maneuverability of rotary-wing flight with the efficiency of fixed-wing flight. The technique to improve FWUAV flight time, presented in this work incorporates (1) the modeling of Robo Raven’s flapping/gliding potential through the development of a state-space representation directly linking Robo Raven’s onboard battery dynamics with its aerodynamic performance, (2) the use of the state-space model to characterize the effect of intermittent gliding techniques on flight performance through simulation, (3) the real-world characterization of the simulation and of intermittent gliding techniques through flight demonstrations, and (4) the development of a design space by which the effect of wing design on gliding performance might be explored and lead to the potential tailoring of wing design to desired flight performance. The expected outcome of this technique is scientific analysis of the extension of Robo Raven’s flight time without added complexity of weight of the battery system.Item COCHLEAR IMPLANTATION: PATH PLANNING ALGORITHMS AND DYNAMICS(2022) Poley, Celeste; Balachandran, Balakumar; Krieger, Axel; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The focus of this dissertation is on incorporating robotics into pediatric cochlear implantation surgery. Since the 1980s, over 300,000 cochlear implantation surgeries have been performed worldwide, both in adults and children alike. For this dissertation research, surgical constraints in the operating theater are of utmost importance for the health and safety of the patient. As the field moves toward minimally invasive surgery, the issues that come with this, such as the loss of the natural field of view and the loss of tactile sense can create significant hurdles for surgeons. Medical robotics can be used to decrease the limitations of such surgical procedures since a desirable attribute of surgical robots is dexterity. Medical robotics can be used to can be used to counter these limitations, by taking advantage of the dexterity of surgical robots. These robots can be used in complex working environments for surgical procedures such as cochlear implantation surgery (CIS). The author's dissertation contains simulation, analytical, and numerical research, through which the effects of dynamics within the path planning algorithms on simulated and modeled cochlear implantation surgery have been studied. A novel path planning algorithm has been developed by making use of Rapidly-exploring Randomized Trees (RRT), and subsequently incorporating Sequential Quadratic Programming. The goal in utilizing a path planning algorithm (PPA) would be to increase safety and aid surgeons in a tightly constrained environment of pediatric temporal bone, which differs in geometry and size from adult temporal bones, and to positively impact the surgical procedure and recuperation from surgery. This algorithm was chosen for use in the tight spaces presented by pediatric patient anatomy and to address patient specific constraints or abnormalities that arise with cochlear implantation surgery in cases of congenital deafness, to add an extra layer of safety for patients. This method allows for more torque handling in tiny and heavily constrained environments, and prevents nicking of delicate anatomy, such as the cranial facial nerve, which can cause facial muscle paralysis if exposed to the slightest damage. The testing of the planning algorithm is carried out in a programming environment. These models are based on geometric equations describing the inner ear anatomy, based on data collected as a part of this dissertation work. Through this doctoral dissertation research, the author has developed several novel methods and innovative techniques to improve associated path planning algorithm. Since cochlear implantation surgeries are moving in the direction of minimally invasive surgery, it would be a beneficial goal to improve the surgery by including a path planning algorithm and a simulated robotic system to help reach the dissertation's goal of simulating the drilling done for cochlear implantation surgery, and with the ultimate goal of improving patient outcome and minimizing time to recovery.Item INVESTIGATING FLUIDIC ENHANCEMENTS FOR SOFT ROBOTIC APPLICATIONS(2021) Acevedo, Ruben; Sochol, Ryan D; Bruck , Hugh A; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Over the past decade, the field of soft robotics has established itself as uniquely suited for applications that would be difficult or impossible to realize using traditional, rigid robots. However, soft robotic systems suffer from two limitations: (i) the inability for soft robots to withstand and transfer high forces and (ii) the tyranny of interconnects for in which each individual fluidic soft actuator either requires its own power source or for the input fluid to be regulated by external electronic valves. In this dissertation, we investigated how to fluidically enhance soft robotic systems to reduce their inherent limitations through the use of negative pressure via layer jamming for programmable variable stiffness and fluidic control via microfluidic circuitry. More specifically, we investigate the use of layer jamming to enhance soft robotic capabilities in (i) a multifunctional sail, (ii) a soft/rigid hybrid robot, and (iii) a multimode actuator and studied the effects layer decohesion has on the mechanical response of layer jamming composites. We also investigated the efficacy of a PolyJet multi-material additive manufacturing strategy to fabricate complete soft robots with fully integrated microfluidic circuitry components such as microfluidic diodes, capacitors, and transistors under three fluidic analogues of conventional electronic signals: (i) constant-flow (i.e., “direct current (DC)”) input conditions, (ii) “alternating current (AC)”-inspired sinusoidal conditions, and (iii) a preprogrammed aperiodic (“variable current”) input. Having fluidically enhanced soft robotic systems will eliminate the need for electronic valves and processors while enable the capability of withstanding and transferring forces found in normal day to day activities, to accelerate their adaptation into mainstream applications. The work to reduce the inherent disadvantages of soft robotic systems offers unique promise to enable new classes of soft robots.Item Numerical and Experimental Studies on Dynamic Interactions of Robot Appendages with Granular Media(2021) Ravula, Preethi; Balachandran, Balakumar; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Terramechanics plays an important role in the design and control of robots moving on granular surfaces. Traction capabilities, slippage, and sinkage of a robot are governed by the interaction of a robot's appendage (such as wheel, track or leg) with the operating terrain and how the terrain motion happens with respect to the appendage during such an interaction. In this dissertation work, dynamics of robot appendages interaction with granular media is explored through numerical and experimental studies. A two dimensional (2D) numerical model, constructed using the Discrete Element Method (DEM), is adapted to simulate lugged wheel interaction with granular media. Parametric studies on wheel performance are conducted for two different control schemes, namely, a slip-based control scheme and an angular velocity-based wheel control scheme. Furthermore, the soil flow pattern under the wheel is studied by examining the force distribution and evolution of force networks during the course of wheel travel.An experimental setup is designed to study the particle motion and force networks inside the media during dynamic forcing. Two different designs of robot appendages, a lugged and a single actuator pendulum are investigated. High speed imaging of photo-elastic particles under polarized light is used to visualize the force distributions inside the media. Qualitative behavior of force chains/networks evolution during interaction with the lugged wheel and pendulum is presented. In addition, quantitative measures of the interaction between appendage and granular media, such as, the driving torque values, appendage velocity, and particle motion are inferred from the experimental findings. Based on this work, insights can be gained into the design influences of robot appendages on performance and further understanding can be obtained on the behavior of granular media across different length scales. Furthermore, the numerical and experimental techniques developed and outcomes of this dissertation can serve as an important foundation for optimal design and control of different robot appendages interacting with deformable surfaces.Item HIGH WAVE VECTOR ACOUSTIC METAMATERIALS: FUNDAMENTAL STUDIES AND APPLICATIONS(2020) Ganye, Randy Tah; Yu, Miao; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Acoustic metamaterials are artificially engineered structures with subwavelength unit cells that hold extraordinary acoustic properties. Their ability to manipulate acoustic waves in ways that are not readily possible in naturally occurring materials have garnered much attention by researchers in recent years. In this dissertation work, acoustic metamaterials that enable wave propagation with high wave vector values are studied. These materials render several key properties, including energy confinement and transport, wave control enhancement, and enhancement of acoustic radiation, which are exploited for enhancing acoustic wave emission and reception. The dissertation work is summarized as follows. First, to enable experimental studies of the deep subwavelength cavities in these metamaterials, a low dimensional fiber optic probe was developed, which allows direct characterization of the intrinsic properties of the metamaterials without seriously disrupting the acoustic fields. Second, low dimensional acoustic metamaterials for enhancing acoustic reception were realized and studied. These metamaterials were demonstrated to achieve both passive and active functionalities, including passive signal amplification and frequency filtering, as well as active tuning for switching and pulse retardation control. Third, a metamaterial emitter was realized and studied, which is capable of enhancing the radiative properties of an embedded emitter. Parametric studies enhanced the understanding of the effects of different geometric parameters on the radiation performance of the structure. Finally, the metamaterial emitter and receiver were combined to form a metamaterial-based sonar system. For the first time, the superior performance of the metamaterial enhanced sonar system over conventional sonar systems was analytically and experimentally demonstrated. As a proof of concept, a robotic sonar platform equipped with the metamaterial system was shown to possess remarkably better tracking performance compared to the conventional system. Through this dissertation work, an enhanced understanding of high-k acoustic metamaterials has been achieved, and their applications in acoustic sensing, emission enhancement, and sonar systems have been demonstrated.Item MULTI-VEHICLE ROUTE PLANNING FOR CENTRALIZED AND DECENTRALIZED SYSTEMS(2019) Patel, Ruchir; Herrmann, Jeffrey W; Azarm, Shapour; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Multi-vehicle route planning is the problem of determining routes for a set of vehicles to visit a set of locations of interest. In this thesis, we describe a study of a classical multi-vehicle route planning problem which compared existing solutions methods on min-sum (minimizing total distance traveled) and min-max (minimizing maximum distance traveled) cost objectives. We then extended the work in this study by adapting approaches tested to generate robust solutions to a failure-robust multi vehicle route planning problem in which a potential vehicle failure may require modifying the solution, which could increase costs. Additionally, we considered a decentralized extension to the multi-vehicle route planning problem, also known as the decentralized task allocation problem. The results of a computational study show that our novel genetic algorithm generated better solutions than existing approaches on larger instances with high communication quality.