No description available.

No description available.

No description available.

No description available.

No description available.

As robots are adopted and deployed in increasingly complex scenarios, simple specifications such as stability and reachability become insufficient to specify the desired behaviors of robots. Temporal logic provides a mathematical formalism for specifying complex, time-related rules. Hence, control synthesis under temporal logic specifications has received significant interest recently. This thesis focuses on a widely used logic in robotics called Signal Temporal Logic (STL), which is defined over real-valued signals. STL is equipped with both qualitative semantics, which shows whether a specification is satisfied, and quantitative semantics (also known as robustness), which measures how strongly a specification is satisfied. Taking advantage of the robustness, control synthesis from STL specifications can be formulated as an optimization problem. Traditional solutions, such as mixed integer programs and gradient-based methods, are computationally expensive (preventing real-time control), model-based (requiring the system model to be known), and centralized (for multi-agent systems). In this thesis, we study the use of machine learning methods in STL control synthesis problems to solve the above limitations. We state our contributions in two core areas: single-agent control and multi-agent coordination. For single-agent scenarios, our first contribution is to parameterize the control policy as a Recurrent Neural Network (RNN) so that the control depends not only on the current system state but also on the history states, which is necessary in general to satisfy STL specifications. Two training strategies for the RNN controller are proposed. The first is an imitation learning approach, where a dataset containing satisfying trajectories is generated, and then the RNN controller is trained on this dataset. The second is a Reinforcement Learning (RL) approach, where the system model is unknown and learned together with the control policy with no need for a dataset. Although these two approaches achieve very high satisfaction according to our simulations and experiments, there is no formal guarantee that the RNN controller can satisfy the specifications. Hence, we propose the third approach, where time-varying High Order Control Barrier Functions (HOCBFs) are constructed from the STL specification and integrated into the RNN controller to guarantee its correctness. Finally, in the case that the specification is not given and only a set of expert demonstrations is available, a generative adversarial imitation learning approach is proposed to simultaneously learn an STL formula describing the underlying requirements followed by the expert and an RNN control policy that satisfies these rules. Since many real-world tasks require the collaboration of teams of robots to finish, we extend the above approaches to multi-agent coordination. We first design a novel logic called Capability Temporal Logic plus (CaTL+). CaTL+ has a two-layer STL structure designed to specify behaviors for heterogeneous teams of robots, which is more efficient and scalable than standard STL, especially when the team is large. Second, we propose a neural network framework called CatlNet to learn both the distributed control policies and communication strategies under CaTL+ specifications, showing good scalability for large robotic teams. / 2026-05-23T00:00:00Z

This research dissertation addresses the significant challenge of Multi-Robot Multi-Target Tracking (MR-MTT), a critical system in various scenarios, including search-and-rescue missions, surveillance, and environmental monitoring. MR-MTT involves coordinating a team of robots to track multiple dynamic targets in diverse environments. This challenge requires efficient coordination among the robots to ensure effective tracking of all targets. The core of this challenge lies in developing efficient strategies for estimation, communication, and control within these robotic systems. Our goal is to create and test different solutions to this general problem. A significant focus of this research is on the estimation aspect of MR-MTT. The system employs a novel distributed Multiple Hypothesis Tracker (MHT) for accurate estimation of both the number and states of multiple targets. A standout feature of our methodology is the introduction of an innovative data association method, designed to reallocate target tracks among the robots, thereby enhancing the collective tracking accuracy and efficiency of the team. This approach is particularly beneficial in scenarios with numerous targets and highly dynamic movements, as it allows for more flexible and responsive tracking. In addition to estimation, a substantial portion of this research focuses on the development of advanced control strategies to enhance the team's efficiency in locating all targets and achieving this goal more swiftly. We have integrated Particle Swarm Optimization (PSO), Simulated Annealing (SA), Ant Colony Optimization (ACO) and Artificial Immune System (AIS)into the MR-MTT systems with an estimation method using the PHD filter. This integration aims to optimize the robots' trajectories and search patterns, leveraging the strengths of these metaheuristic-based algorithms to strike a balance between exploration and exploitation. Such optimization is crucial for enhancing the overall efficiency and effectiveness of MR-MTT systems. Furthermore, this dissertation includes a comprehensive system-level analysis and prediction component through dimensionless variable analysis. We have developed an analytical framework to evaluate and predict the performance of MR-MTT systems under various operational scenarios and environmental conditions. This framework is intended to provide deep insights into the critical performance determinants and their interrelations, guiding the design and optimization of MR-MTT systems. The anticipated outcomes of this research include improved accuracy in target tracking, enhanced performance metrics for MR-MTT systems, and valuable insights for the future design and management of sophisticated multi-robot systems. The expected outcomes of this research are multifold: enhanced accuracy in target tracking, improved performance metrics for MR-MTT systems, and valuable insights for the future development and management of complex multi-robot systems. Through this proposal, we aim to contribute significantly to the field of robotic coordination and tracking, addressing critical needs in various high-impact applications. / Mechanical Engineering

Thesis(M.S.)--Auburn University, 2005. / Abstract. Vita. Includes program. Includes bibliographic references.

Stackable balloon actuators (SBAs) present a compelling new actuator for healthcare and scientific exploration applications. The ability of these multi degree-of-freedom actuators to perform large deformations and conform to the shape of objects make them a valuable choice for grasping and manipulation tasks. This work explores how the design of the SBA can be manipulated through a bio-inspired lens to exploit the features of an octopus arm. It begins by manipulating the design of the SBAs, taking inspiration from the octopus arm morphology to explore different layer geometries, variable layer sizes to generate a tapered profile, embedding different materials to improve force output, and stacking individually actuated SBAs in series to produce a more dexterous assembly. These design manipulations are then characterized individually and as components of multi-arm systems to determine their performance in grasping and object manipulation tasks. / 2027-05-31T00:00:00Z

Mechanical compliance will be an essential component for agile robots as they begin to leave the laboratory settings and join our world. The most crucial finding of this dissertation is showing how lessons learned from soft robotics can be adapted into traditional robotics to introduce compliance. Therefore, it presents practical knowledge on how to build soft bodied sensor and actuation modules: first example being soft-bodied curvature sensors. These sensors contain both standard electronic components soldered on flexible PCBs and hyperelastic materials that cover the electronics. They are built by curing multi-material composites inside hyper elastic materials. Then it shows, via precise sensing by using magnets and Hall-effect sensors, how closed-loop control of soft actuation modules can be achieved via proprioceptive feedback. Once curvature sensing idea is verified, the dissertation describes how the same sensing methodology, along with the same multi-material manufacturing technique can be utilized to construct soft bodied tri-axial force sensors. It shows experimentally that these sensors can be used by traditional robotic grippers to increase grasping quality. At this point, I observe that compliance is an important property that robots may utilize for different types of motions. One example being Raibert's 2D hopper mechanism. It uses its leg-spring to store energy while on the ground and release this energy before jumping. I observe that via soft material design, it would be possible to embed compliance directly into the linkage design itself. So I go over the design details of an extremely lightweight compliant five-bar mechanism design that can store energy when compressed via soft ligaments embedded in its joints. I experimentally show that the compliant leg design offers increased efficiency compared to a rigid counterpart. I also utilize the previously mentioned soft bodied force sensors for rapid contact detection (~5-10 Hz) in the hopper test platform. In the end, this thesis connects soft robotics with the traditional body of robotic knowledge in two aspects: a) I show that manufacturing techniques we use for soft bodied sensor/actuator designs can be utilized for creating soft ligaments that add strength and compliance to robot joints; and b) I demonstrate that soft bodied force sensing techniques can be used reliably for robotic contact detection.