Robotics

Since 2010, aquaculture practices have produced 70% of global seafood consumption. However, this fast-growing sector of agriculture has yet to see the adoption of advanced technologies to improve farm operations. The Hybrid Aerial Underwater robotiCs System (HAUCS) is an Internet of Things (IoT) framework that aims to bring transformative changes to pond aquaculture.
This project focuses on the latest developments in the HAUCS mobile sensing platform and field deployment. A novel rigid Kirigami-based robotic extension subsystem was created to expand the functionality of the HAUCS platform. The primary objective of this design was to limit the surface area of an extender arm on the drone during flight operations and minimize the in-flight drag. By utilizing a novel combination of shape memory polymer (SMP) and nitinol to extend and retrieve the sensing arm, the structure was able to conserve energy while operating under varying environmental conditions.

The sensation of touch is an integral part of using our hands. As different researchers work toward the restoration of afferent sensation in prosthetic hands, it becomes urgent to better understand how an artificial hand’s afferent inputs are affected by the efferent muscular outputs, and vice-versa. Current methods of neuroprosthetic research have many regulatory hurdles, time, cost, and associated risk to the patient. To circumvent these hurdles, we developed a non-invasive, closed-loop (CL) neuroprosthetic research platform, integrating artificial tactile signals from an artificial hand with biomimetically-stimulated biological neuronal networks (BNNs) cultured in a multielectrode array (MEA) chamber. These living embodied biological computers (EBCs) can provide a non-invasive alternative for investigating invasive neuroprosthetic interfaces. With them we can explore a variety of control techniques, tactile sensation encoding methods, and neural decoding methods to increase the rate of research in this area with minimal regulatory approval, greatly reduced cost and time, and no risk to the patients. In the first stage of this integration, our EBC was programmed to embody neuronal spiking from spontaneously active “efferent” receptive fields in cultured BNNs as intentional signals for movement. Bursts were transferred to a robotic hand and initiated a tapping motion of the index finger laid in proximity to a surface. Contact elicited artificial sensations, which were registered by a biotac tactile sensor array fit to the robotic fingertip.

Urban freight system constitutes an essential component for both economic and social aspects of the urban areas. However, the driving forces of globalization and ecommerce have adversely affected the volume of freight vehicles in urban roads over the past few decades impacting the sustainability and efficiency of last-mile deliveries. At the same time, the last-mile problem of goods distribution from companies to customers comprises one of the most costly and highest polluting components of the supply chain. Over the past few years, different innovative concepts of autonomous vehicles were introduced to improve last-mile logistic inefficiencies such as traffic congestion and pollution externalities. The objective of this study is to optimize a distribution network consisting of a set of depots and customers by utilizing Autonomous Delivery Robots (ADRs). For that reason, a Mixed Integer Linear Programming model was developed in GAMS for solving the vehicle routing problem while minimizing the total delivery and delay costs of ADRs. This optimization model is based on the route assignment and the required number of ADRs within the network. A heuristic solution algorithm based on the cluster-first, route-second technique was developed in MATLAB for solving the NP-hard problem efficiently. First the customers were clustered to depots based on their maximum distance from them and the maximum allowed number of customers per cluster. After the clustering, the mathematical model was implemented in each cluster providing an exact solution. Three different medium-sized scenarios of 200, 300 and 400 customers were tested under three different clustering instances of a maximum of 20, 30 and 40 customers per cluster and their results were presented and discussed in detail.

As artificial intelligence (AI), such as reinforcement learning (RL), has continued to grow, the introduction of AI for use in robotic arms in order to have them autonomously complete tasks has become an increasingly popular topic. Robotic arms have recently had a drastic spike in innovation, with new robotic arms being developed for a variety of tasks both menial and complicated. One robotic arm recently developed for everyday use in close proximity to the user is the Kinova Gen 3 Lite, but limited formal research has been conducted about controlling this robotic arm both with an AI and in general. Therefore, this thesis covers the implementation of Python programs in controlling the robotic arm physically as well as the use of a simulation to train an RL based AI compatible with the Kinova Gen 3 Lite. Additionally, the purpose of this research is to identify and solve the difficulties in the physical instance and the simulation as well as the impact of the learning parameters on the robotic arm AI. Similarly, the issues in connecting two Kinova Gen 3 Lites to one computer at once are also examined.
This thesis goes into detail about the goal of the Python programs created to move the physical robotic arm as well as the overall setup and goal of the robotic arm simulation for the RL method. In particular, the Python programs for the physical robotic arm pick up the object and place it at a different location, identifying a method to prevent the gripper from crushing an object without a tactile sensor in the process. The thesis also covers the effect of various learning parameters on the accuracy and steps to goal curves of an RL method designed to make a Kinova Gen 3 Lite grab an object in a simulation. In particular, a neural network implementation of RL method with one of the learning parameters changed in comparison to the optimal learning parameters. The neural network is trained using Python Anaconda to control a Kinova Gen 3 Lite robotic arm model for a simulation made in the Unity compiler.

As technology progresses, tasks involving object manipulation that were once conducted by humans are now being accomplished through robots. Specifically, robots carry out these goals through the utilization of different forms of artificial intelligence, including deep learning via a convolutional neural network. One robot made to accomplish this purpose is the ROS controlled TurtleBot3 Waffle Pi with an OpenMANIPULATOR-X robotic arm. This type of TurtleBot3 was developed with the express purpose of education and research but may not be limited to those two usages. Based on the current design of this classification of TurtleBot3, it may have multiple applications outside the testing environment, granting it further uses in a variety of tasks. The TurtleBot3 is easy to setup to fulfill the purposes for which the TurtleBot3 Waffle Pi was designed, and the exploration into further uses would allow for the discovery of alternatives to some tasks that normally require more work. For that reason, this thesis was conducted to determine the various uses of the TurtleBot3 with a robotic arm and if this robot can be used outside of a testing environment for various real-world tasks.

Robotics have advanced to include highly anthropomorphic (human-like) entities. A novel eye-tracking paradigm was developed to assess infants’ sensitivity to communicative gestures by human and robotic informants. Infants from two age groups (5-9 months, n = 25; 10-15 months, n = 9) viewed a robotic or human informant pointing to locations where events would occur during experimental trials. Trials consisted of three phases: gesture, prediction, and event. Duration of looking (ms) to two areas of interest, target location and non-target location, was extracted. A series of paired t-tests revealed that only older infants in the human condition looked significantly longer to the target location during the prediction phase (p = .036). Future research is needed to tease apart what components of the robotic hand infants respond to differentially, and whether a robotic hand can be manipulated to increase infants’ sensitivity to social communication gestures executed by said robotic hand.

Different innovative concepts are aiming to improve last-mile urban logistics and reduce traffic congestion. Congested metropolitan cities are implementing last-mile delivery robots to make the delivery cheaper and faster. A key factor for the success of Automated Delivery Robots (ADRs) in the last-mile is its ability to meet the fluctuating demand for robots at each micro-hub. Delivery companies rent robots from micro-hubs scattered around the city, use them for deliveries, and return them at micro-hubs. This paper studies the dynamic assignment of the robots to satisfy their demands between the micro-hubs. A Mixed-Integer Linear Programming (MILP) model is developed, which minimizes the total transportation costs by determining the optimum required fleet size. The result determines the number of robots required for each planning period to meet all the demands. It provides algorithms to operate and schedule the robot-sharing system in the last leg of the delivery in dense urban areas.

This thesis presents the work done to deliver a robotic system that provides assistance to operators at nuclear waste cleaning facilities. The work done to deliver such system was focused on robotic control and tactile sensing abilities. Haptic feedback mechanism was also added to the system to convey information for the operator. First chapter of the thesis introduces the goals and objectives of this project as well as a detailed literature review on the subsystems used. Second chapter presents previous work done in the area of soft robotics. Such work proved important as the haptic feedback mechanism utilizes a soft robotic armband. Third chapter introduces phase one of the main project. This chapter justifies the use of the selected robots and introduces the concept of adding tactile abilities to the robotic hand used. Chapter four introduces phase two of the project that focused on improving phase one system via a new tactile sensor.

Traditional robots are made from hard materials like hard plastic or metal and consist of
regular rigid mechanical parts. Using those parts has some limitations, like limited
dexterity and lack of flexibility. Some of these limitations could be avoided through using
a compliant material, because it has higher flexibility and dexterity. It is also safer to be in
direct contact with humans. This thesis studies soft pneumatic manipulators (SPMs) that
move in multi degrees of freedom (MDOF), which makes them able to perform various
functions. The study will include designing, fabricating, and testing three different SPMs
with different taper angles -- 0^0, 1^0, and 2^0 -- to measure the effect of varying this geometry
on the achievable force by the end effector and the range of bending and elongation. Every
single SPM consists of three soft pneumatic chambers to reach unlimited points on its
workspace through implementing bending and elongating movements. There are a lot of
applications for this kind of soft actuators, like rehabilitation, underwater utilizes, and
robots for surgery and rescues. Most soft pneumatic actuators provide one kind of movement, for bending, twisting, or elongating. Combining more than one kind of
movement in one soft pneumatic actuator provides considerable contributions to the body
of research. The SPMs were controlled and tested to evaluate the achieved force and two
kinds of movement, bending and elongating range. The results of each module has been
compared with the others to determine which actuator has the best performance. Then a
force controller was created to maintain the desired force that was achieved by the end
effector. The results indicated that the optimal angle of the SPM was 2^0.

The design and construction of a tri-cable, planar robotic device for use in neurophysical rehabilitation is presented. The criteria for this system are based primarily on marketability factors, rather than ideal models or mathematical outcomes. The device is designed to be low cost and sufficiently safe for a somewhat disabled individual to use unsupervised at home, as well as in a therapist's office. The key features are the use of a barrier that inhibits the user from coming into contact with the cables as well as a "break-away" joystick that the user utilizes to perform the rehabilitation tasks. In addition, this device is portable, aesthetically acceptable and easy to operate. Other uses of this system include sports therapy, virtual reality and teleoperation of remote devices.