Robotics Commons^™

Swarm Of One: Self-Organization In Multicellular Robots, Trevor Riley Smith

Graduate Theses, Dissertations, and Problem Reports (ETD)

Designing robots traditionally relies on centralized, top-down methods that constrain adaptability and resiliency, producing systems optimized for specific tasks. However, such systems often fail in complex environments, such as extraterrestrial, disaster, or natural settings, where designers cannot foresee every scenario a robot may encounter. Achieving autonomy in these contexts requires robots that are resilient and capable of adapting to the unknown. Multicellular organisms exhibit this adaptability through self-organization, where complex structures and behaviors emerge from local interactions among their cells. Such organisms can be viewed simultaneously as a single entity and a swarm of cooperating cells, i.e., a Swarm-of-One.

This …

Semantic-Weighted Lidar–Camera Slam For Robust Mapping In Vegetation-Rich Environments, David Aigbovboise Akhihiero

Graduate Theses, Dissertations, and Problem Reports (ETD)

Many LiDAR odometry and SLAM methods are lightweight and effective, but they assume that all LiDAR point returns are equally reliable. In practice, this assumption is often violated in vegetation rich or semi transparent environments, where in a single beam footprint, LiDAR beams may intersect multiple scattering surfaces like leaves, branches or glass. These ambiguities in returns produces probabilistic or biased range measurements that violate the assumptions of most geometric registration frameworks and can lead to drift and inconsistency.

To address this challenge, this dissertation proposes a semantic-weighted LiDAR-camera SLAM framework that integrates class dependent weights into a front-end LiDAR …

Efficient And Test-Time Adaptive Visual Object Tracking In The Wild, Ram J. Zaveri

Graduate Theses, Dissertations, and Problem Reports (ETD)

Tracking a single object, given the location at the first frame, has been an ongoing challenge in the vision community for decades. Most recent approaches provide reasonably good performance, especially when benchmarked on in-distribution (ID) datasets, i.e., on the testing portion of the same datasets used for training. However, they incur high computational costs and hardware constraints, making their deployment in the wild for mobile, autonomous, and IoT applications still challenging. Efficient visual trackers address the efficiency aspect of such bottlenecks; however, they tend to overfit to their training distributions and lack generalization abilities, resulting in them performing well on …

Imitation Learning In Robotic Manipulation Using Diffusion Models, Marlon Domingues De Oliveira

Graduate Theses, Dissertations, and Problem Reports (ETD)

This work deals with the problem of teaching robots by demonstration, also known by imitation learning. The main objective of this area of research is to enable the autonomous execution of complex robotic tasks using neural networks trained on expert-generated data, thus allowing the transfer of human knowledge to machines. To this end, this thesis describes an experimental setup especially designed for the study of imitation learning in robotic manipulation tasks and the adaptation and evaluation of a previously published technique to this setup. The experimental setup developed in this work is based on the Franka Emika Research 3 manipulator, …

Flower Cluster Matching Utilizing The Unscented Transform For Robotic Pollination, Andy Chu

Graduate Theses, Dissertations, and Problem Reports (ETD)

The use of automated systems for agriculture is integral to keeping the food supply secure. Both industry and academia are exploring and applying methods to increase the yield of plants in environments ranging from outdoor fields to greenhouses. Specifically, many automated systems use continuous monitoring of plants to track plant health and yield. The use of computer vision is necessary when it comes to precision operations that use robotics. Today, robots are trained to weed, harvest, and pollinate. To accomplish these tasks autonomously, a lot of data is needed, which is where spatial-temporal observations of the plants are being recorded …

A Framework For Biomimetic Robot Design Applied To The Development Of A Robotic Model Of Drosophila Melanogaster, Clarissa A. Goldsmith

Graduate Theses, Dissertations, and Problem Reports (ETD)

For decades, the field of biologically inspired robotics has leveraged insights from animal locomotion to improve the walking ability of legged robots. Recently, “biomimetic” robots have been developed to model how specific animals walk. By prioritizing biological accuracy to the target organism rather than the application of general principles from biology, these robots can be used to develop detailed biological hypotheses for animal experiments, ultimately improving our understanding of the biological control of legs while improving technical solutions. Much of this work involves biologically inspired walking controllers informed by the morphology and dynamics of the insect nervous system, which necessitate …

Autonomous Object Search Planning In Large-Scale Environments, Matthew A. Collins

Graduate Theses, Dissertations, and Problem Reports (ETD)

The advancement of autonomous search holds significant promise for applications ranging from emergency response to planetary exploration. This thesis investigates strategies to enhance autonomous search performance in large-scale environments. The main contribution of this work is its practical application in real-world scenarios, where efficient search methods are essential for managing vast amounts of data, particularly in large environments. Effective search planning requires navigating complexities such as limited prior information and managing large state spaces, necessitating advanced strategies to plan with this limited information. Additionally, balancing exploration and exploitation is crucial for optimizing the search process, as it ensures thorough coverage …

On Uncertainty For Ill-Posed Robot Decision Problems, Jared Joseph Beard

Graduate Theses, Dissertations, and Problem Reports (ETD)

As robots adopt more real world responsibilities, they will be expected to solve more complicated problems. In some cases limited prior knowledge will result in unmodelled environmental conditions; in others, multiple users may have competing perspectives on how to frame a decision problem. Many existing frameworks, namely Markov decision processes (MDP) presuppose users have identified a specific problem with models sufficient to solve or learn a problem. If we wish to extend MDPs to novel problems or those heavily dependent on user feedback, autonomous decision makers must be able to identify limitations in how a given problem is framed and …

An Approach For Robotic Pollination That Utilizes Imitation Learning, Ronald Michael Butts Ii

Graduate Theses, Dissertations, and Problem Reports (ETD)

The global decline in pollinator populations poses a significant threat to agriculture, motivating the development of robotic pollination systems. Previous works demonstrated successful robotic pollination of bramble flowers using visual servoing; however, pollination was limited to specific flower orientations. As such, the objective of this work is to develop a robotic pollination system that is capable of pollinating a wider range of orientations.

This research introduces an imitation learning-based framework for robotic pollination that positions the manipulator to view chosen flowers in specific orientations. The developed model leverages object detection (YOLOv8) to identify individual flowers and a convolutional neural network …

Enhancing Robotic Exploration Through Semantically-Guided Sampling Strategies, Christopher Alexander Arend Tatsch

Graduate Theses, Dissertations, and Problem Reports (ETD)

From space and deep-sea exploration to disaster response and environmental monitoring, autonomous robots are essential for advancing science, improving safety, and addressing critical challenges. This dissertation introduces a novel open-source strategy for autonomous robotic exploration: the Semantically-Guided Exploration (SGE) framework. Designed for ground vehicles, SGE integrates semantic understanding into the autonomous exploration process, improving decision-making in complex environments. Specifically, the proposed sampling-based approach uses the information from the semantic segmentation of RGB images and depth images to guide the robot's selection of exploration goals. This method enables the robot to steer away from potential dangers such as large rocks and …

Foraging Economies: A Market Based Methodology For Robotic Swarm Foraging, John A. Little

Graduate Theses, Dissertations, and Problem Reports (ETD)

Swarm robotics involves coordinating large groups of autonomous agents to accomplish complex tasks through decentralized, adaptive behaviors, providing a robust and scalable approach suited to dynamic and unpredictable environments. While traditional swarm models frequently draw inspiration from biological systems such as ant colonies or bee foraging, other approaches use techniques from physics, control theory, and economics to achieve effective coordination. This study distinguishes itself by applying economic principles—specifically, market-driven mechanisms like auctions, utility functions based on opportunity cost, and supply-demand dynamics based on fluctuating resource values at a central base—to improve task allocation within a swarm foraging context. This approach …

Joint Learning Of Unknown Safety Constraints And Control Policies In Reinforcement Learning, Lunet Abiye Yifru

Graduate Theses, Dissertations, and Problem Reports (ETD)

Reinforcement learning (RL) has revolutionized decision-making across a wide range of domains over the past few decades. Yet, deploying RL policies in real-world scenarios presents the crucial challenge of ensuring safety. Traditional safe RL approaches have predominantly focused on incorporating predefined safety constraints into the policy learning process. However, this reliance on predefined safety constraints poses limitations in dynamic and unpredictable real-world settings where such constraints may not be available or sufficiently adaptable. Bridging this gap, we propose a novel approach that concurrently learns a safe RL control policy and identifies the unknown safety constraint parameters of a given environment. …

Motion Planning In Artificial And Natural Vector Fields, Bernardo Martinez Rocamora Junior

Graduate Theses, Dissertations, and Problem Reports (ETD)

This dissertation advances the field of autonomous vehicle motion planning in various challenging environments, ranging from flows and planetary atmospheres to cluttered real-world scenarios. By addressing the challenge of navigating environmental flows, this work introduces the Flow-Aware Fast Marching Tree algorithm (FlowFMT*). This algorithm optimizes motion planning for unmanned vehicles, such as UAVs and AUVs, navigating in tridimensional static flows. By considering reachability constraints caused by vehicle and flow dynamics, flow-aware neighborhood sets are found and used to reduce the number of calls to the cost function. The method computes feasible and optimal trajectories from start to goal in challenging …

Imitation Learning For Swarm Control Using Variational Inference, Hafeez Olafisayo Jimoh

Graduate Theses, Dissertations, and Problem Reports (ETD)

Swarms are groups of robots that can coordinate, cooperate, and communicate to achieve tasks that may be impossible for a single robot. These systems exhibit complex dynamical behavior, similar to those observed in physics, neuroscience, finance, biology, social and communication networks, etc. For instance, in Biology, schools of fish, swarm of bacteria, colony of termites exhibit flocking behavior to achieve simple and complex tasks. Modeling the dynamics of flocking in animals is challenging as we usually do not have full knowledge of the dynamics of the system and how individual agent interact. The environment of swarms is also very noisy …

Exploiting The Advantages And Overcoming The Challenges Of The Cable In A Tethered Drone System, Rogerio Rodrigues Lima

Graduate Theses, Dissertations, and Problem Reports (ETD)

This dissertation proposes solutions for motion planning, localization, and landing of tethered drones using only tether variables. A tether-based multi-model localization framework for tethered drones is proposed. This framework comprises three independent localization strategies based on a different model. The first strategy uses simple trigonometric relations assuming that the tether is taut; the second method relies on a set of catenary equations for the slack tether case; the third estimator is a neural network-based predictor that can cover different tether shapes. Multi-layer perceptron networks previously trained with a dataset comprised of the tether variables (i.e., length, tether angles on the …

System Development Of An Unmanned Ground Vehicle And Implementation Of An Autonomous Navigation Module In A Mine Environment, Jonas Amoama Bredu Jnr

Graduate Theses, Dissertations, and Problem Reports (ETD)

There are numerous benefits to the insights gained from the exploration and exploitation of underground mines. There are also great risks and challenges involved, such as accidents that have claimed many lives. To avoid these accidents, inspections of the large mines were carried out by the miners, which is not always economically feasible and puts the safety of the inspectors at risk. Despite the progress in the development of robotic systems, autonomous navigation, localization and mapping algorithms, these environments remain particularly demanding for these systems. The successful implementation of the autonomous unmanned system will allow mine workers to autonomously determine …

Improving Robotic Decision-Making In Unmodeled Situations, Nicholas Scott Ohi

Graduate Theses, Dissertations, and Problem Reports (ETD)

Existing methods of autonomous robotic decision-making are often fragile when faced with inaccurate or incompletely modeled distributions of uncertainty, also known as ambiguity. While decision-making under ambiguity is a field of study that has been gaining interest, many existing methods tend to be computationally challenging, require many assumptions about the nature of the problem, and often require much prior knowledge. Therefore, they do not scale well to complex real-world problems where fulfilling all of these requirements is often impractical if not impossible. The research described in this dissertation investigates novel approaches to robotic decision-making strategies which are resilient to …

Multimodal Adversarial Learning, Uche Osahor

Graduate Theses, Dissertations, and Problem Reports (ETD)

Deep Convolutional Neural Networks (DCNN) have proven to be an exceptional tool for object recognition, generative modelling, and multi-modal learning in various computer vision applications. However, recent findings have shown that such state-of-the-art models can be easily deceived by inserting slight imperceptible perturbations to key pixels in the input. A good target detection systems can accurately identify targets by localizing their coordinates on the input image of interest. This is ideally achieved by labeling each pixel in an image as a background or a potential target pixel. However, prior research still confirms that such state of the art targets models …

Localization Algorithms For Gnss-Denied And Challenging Environments, Chizhao Yang

Graduate Theses, Dissertations, and Problem Reports (ETD)

In this dissertation, the problem about localization in GNSS-denied and challenging environments is addressed. Specifically, the challenging environments discussed in this dissertation include two different types, environments including only low-resolution features and environments containing moving objects. To achieve accurate pose estimates, the errors are always bounded through matching observations from sensors with surrounding environments. These challenging environments, unfortunately, would bring troubles into matching related methods, such as "fingerprint" matching, and ICP. For instance, in environments with low-resolution features, the on-board sensor measurements could match to multiple positions on a map, which creates ambiguity; in environments with moving objects included, the …

Planetary Rover Inertial Navigation Applications: Pseudo Measurements And Wheel Terrain Interactions, Cagri Kilic

Graduate Theses, Dissertations, and Problem Reports (ETD)

Accurate localization is a critical component of any robotic system. During planetary missions, these systems are often limited by energy sources and slow spacecraft computers. Using proprioceptive localization (e.g., using an inertial measurement unit and wheel encoders) without external aiding is insufficient for accurate localization. This is mainly due to the integrated and unbounded errors of the inertial navigation solutions and the drifted position information from wheel encoders caused by wheel slippage. For this reason, planetary rovers often utilize exteroceptive (e.g., vision-based) sensors. On the one hand, localization with proprioceptive sensors is straightforward, computationally efficient, and continuous. On the other …

Designs And Practical Control Methods For Soft Parallel Robots, Benjamin T. Buzzo

Graduate Theses, Dissertations, and Problem Reports (ETD)

The use of soft robotics is becoming an increasingly researched topic, since they can provide more flexibility in movements and increase safety when working with humans. However, they are more susceptible to modeling and manufacturing errors in the design.

The objective of this thesis is two-fold, the first objective is to determine the benefits and limitations of using calibration tables that rely on the PWM signals instead of modeling as a control method. If calibration tables are not adequate to achieve a high level of precision. The second objective is to determine if using a tethered mobile robot in unison …

Active Localization For Robotic Systems: Algorithms And Cost Metrics, Jared Strader

Graduate Theses, Dissertations, and Problem Reports (ETD)

In the real world, a robotic system must operate in the presence of motion and sensing uncertainty. This is caused by the fact that the motion of a robotic system is stochastic due to disturbances from the environment, and the states are only partially observable due noise in the sensor measurements. As a result, the true state of a robotic system is unknown, and estimation techniques must be used to infer the states from the belief, which is the probability distribution over all possible states. Accordingly, a robotic system must be capable of reasoning about the quality of the belief …

Planning Algorithms Under Uncertainty For A Team Of A Uav And A Ugv For Underground Exploration, Matteo De Petrillo

Graduate Theses, Dissertations, and Problem Reports (ETD)

Robots’ autonomy has been studied for decades in different environments, but only recently, thanks to the advance in technology and interests, robots for underground exploration gained more attention. Due to the many challenges that any robot must face in such harsh environments, this remains an challenging and complex problem to solve.

As technology became cheaper and more accessible, the use of robots for underground ex- ploration increased. One of the main challenges is concerned with robot localization, which is not easily provided by any Global Navigation Services System (GNSS). Many developments have been achieved for indoor mobile ground robots, making …

Route Planning For Long-Term Robotics Missions, Christopher Alexander Arend Tatsch

Graduate Theses, Dissertations, and Problem Reports (ETD)

Many future robotic applications such as the operation in large uncertain environment depend on a more autonomous robot. The robotics long term autonomy presents challenges on how to plan and schedule goal locations across multiple days of mission duration. This is an NP-hard problem that is infeasible to solve for an optimal solution due to the large number of vertices to visit. In some cases the robot hardware constraints also adds the requirement to return to a charging station multiple times in a long term mission. The uncertainties in the robot model and environment require the robot planner to account …

Fast Decision-Making Under Time And Resource Constraints, Kyle Gabriel Lassak

Graduate Theses, Dissertations, and Problem Reports (ETD)

Practical decision makers are inherently limited by computational and memory resources as well as the time available in which to make decisions. To cope with these limitations, humans actively seek methods which limit their resource demands by exploiting structure within the environment and exploiting a coupling between their sensing and actuation to form heuristics for fast decision-making. To date, such behavior has not been replicated in artificial agents. This research explores how heuristics may be incorporated into the decision-making process to quickly make high-quality decisions through the analysis of a prominent case study: the outfielder problem. In the outfielder problem, …

Automated And Standardized Tools For Realistic, Generic Musculoskeletal Model Development, Trevor Rees Moon

Graduate Theses, Dissertations, and Problem Reports (ETD)

Human movement is an instinctive yet challenging task that involves complex interactions between the neuromusculoskeletal system and its interaction with the surrounding environment. One key obstacle in the understanding of human locomotion is the availability and validity of experimental data or computational models. Corresponding measurements describing the relationships of the nervous and musculoskeletal systems and their dynamics are highly variable. Likewise, computational models and musculoskeletal models in particular are vitally dependent on these measurements to define model behavior and mechanics. These measurements are often sparse and disparate due to unsystematic data collection containing variable methodologies and reporting conventions. To date, …

Virtual Morphology As A Method Of Robotic Control, Conner Todd Castle

Graduate Theses, Dissertations, and Problem Reports (ETD)

This thesis presents Virtual Morphology (VM), a method that explores a different perspective on the design of robot autonomy using inspiration from morphological computing and programmed computation. Morphological computation offers physical solutions that solve complex tasks, like robotic grasping of unknown objects, with relative ease. Unfortunately, these physical solutions are difficult to adjust post-development, and are usually designed to complete only one or a few specific tasks. Programmed computational approaches are more flexible because they can be implemented and adjusted through software, but unfortunately, these approaches can become rather complex as tasks become more difficult. This thesis explores the potential …

Immunity-Based Framework For Autonomous Flight In Gps-Challenged Environment, Mohanad Al Nuaimi

Graduate Theses, Dissertations, and Problem Reports (ETD)

In this research, the artificial immune system (AIS) paradigm is used for the development of a conceptual framework for autonomous flight when vehicle position and velocity are not available from direct sources such as the global navigation satellite systems or external landmarks and systems. The AIS is expected to provide corrections of velocity and position estimations that are only based on the outputs of onboard inertial measurement units (IMU). The AIS comprises sets of artificial memory cells that simulate the function of memory T- and B-cells in the biological immune system of vertebrates. The innate immune system uses information about …