Background: With the pressing need for increased productivity and delivery of end products of uniform quality, industry is turning more and more to computer-based automation. At the present time, most of industrial automated manufacturing is carried out by specialpurpose machines, designed to perform specific functions in a manufacturing process. The inflexibility and generally high cost of these machines often referred to as hard automation systems, have led to a broad-based interest in the use of robots capable of performing a variety of manufacturing functions in a more flexible working environment and at lower production costs. A robot is a reprogrammable general-purpose manipulator with external sensors that can perform various assembly tasks. A robot may possess intelligence, which is normally due to computer algorithms associated with its controls and sensing systems. Industrial robots are general-purpose, computer-controlled manipulators consisting of several rigid links connected in series by revolute or prismatic joints. Most of today’s industrial robots, though controlled by mini and microcomputers are basically simple positional machines. They execute a given task by playing back a prerecorded or preprogrammed sequence of motion that has been previously guided or taught by the hand-held control teach box. Moreover, these robots are equipped with little or no external sensors for obtaining the information vital to its working environment. As a result robots are used mainly for relatively simple, repetitive tasks. More research effort has been directed in sensory feedback systems, which has resulted in improving the overall performance of the manipulator system. An example of a sensory feedback system would be: a vision Charge-Coupled Device (CCD) system. This can be utilized to manipulate the robot position dependant on the surrounding robot environment (various object profile sizes). This vision system can only be used within the robot movement envelope

<p> Navigating a vehicle autonomously and safely in unknown surroundings to a desired destination is challenging due to lack of initial information about stationary and moving objects along the path. This thesis proposes a navigation system that avoids static and dynamic obstacles using weighted real-time sensor feedback. The effectiveness of the system is demonstrated by implementing it on a robot. A 16-beam solid-state LiDAR sensor is used to detect obstacles to control a differential drive mobile robot. The sensor measurements are weighted and integrated into the Pure Pursuit path following algorithm to avoid obstacles in a natural smooth movement. The primary purpose of this thesis is to integrate all the sensors and processing units to create an appropriate reaction of the robot while it progresses toward the destination. The Algorithm proposed in this work guided the robot safely and fluently from start to end position while avoiding obstacles along the path.</p><p>

abstract: Sports activities have been a cornerstone in the evolution of humankind through the ages from the ancient Roman empire to the Olympics in the 21st century. These activities have been used as a benchmark to evaluate the how humans have progressed through the sands of time. In the 21st century, machines along with the help of powerful computing and relatively new computing paradigms have made a good case for taking up the mantle. Even though machines have been able to perform complex tasks and maneuvers, they have struggled to match the dexterity, coordination, manipulability and acuteness displayed by humans. Bi-manual tasks are more complex and bring in additional variables like coordination into the task making it harder to evaluate. A task capable of demonstrating the above skillset would be a good measure of the progress in the field of robotic technology. Therefore a dual armed robot has been built and taught to handle the ball and make the basket successfully thus demonstrating the capability of using both arms. A combination of machine learning techniques, Reinforcement learning, and Imitation learning has been used along with advanced optimization algorithms to accomplish the task. / Dissertation/Thesis / Masters Thesis Engineering 2016

abstract: Human locomotion is an essential function that enables individuals to lead healthy, independent lives. One important feature of natural walking is the capacity to transition across varying surfaces, enabling an individual to traverse complex terrains while maintaining balance. There has been extensive work regarding improving prostheses' performance in changing walking conditions, but there is still a need to address the transition from rigid to compliant or dynamic surfaces, such as the transition from pavement to long grass or soft sand. This research aims to investigate the mechanisms involved such transitions and identify potential indicators of the anticipated change that can be applied to the control of a powered ankle prosthetic to reduce falls and improve stability in lower-limb amputees in a wider range of walking environments. A series of human subject experiments were conducted using the Variable Stiffness Treadmill (VST) to control walking surface compliance while gait kinematics and muscular activation data were collected from three healthy, nondisabled subjects. Specifically, the kinematics and electromyography (EMG) profiles of the gait cycles immediately preceding and following an expected change in surface compliance were compared to that of normal, rigid surface walking. While the results do not indicate statistical differences in the EMG profiles between the two modes of walking, the muscle activation appears to be qualitatively different from inspection of the data. Additionally, there were promising statistically significant changes in joint angles, especially in observed increases in hip flexion during the swing phases both before and during an expected change in surface. Decreases in ankle flexion immediately before heel strike on the perturbed leg were also observed to occur simultaneously with decreases in tibialis anterior (TA) muscle activation, which encourages additional research investigating potential changes in EMG profiles. Ultimately, more work should be done to make strong conclusions about potential indicators of walking surface transitions, but this research demonstrates the potential of EMG and kinematic data to be used in the control of a powered ankle prosthetic. / Dissertation/Thesis / Masters Thesis Mechanical Engineering 2018

abstract: This thesis presents an autonomous vehicle test bed which can be used to conduct studies on the interaction between human-driven vehicles and autonomous vehicles on the road. The test bed will make use of a fleet of robots which is a microcosm of an autonomous vehicle performing all the vital tasks like lane following, traffic signal obeying and collision avoidance with other vehicles on the road. The robots use real-time image processing and closed-loop control techniques to achieve automation. The testbed also features a manual control mode where a user can choose to control the car with a joystick by viewing a video relayed to the control station. Stochastic rogue vehicle processes will be introduced into the system which will emulate random behaviors in an autonomous vehicle. The test bed was experimented to perform a comparative study of driving capabilities of the miniature self-driving car and a human driver. / Dissertation/Thesis / Masters Thesis Electrical Engineering 2018

abstract: This project is to develop a new method to generate GPS waypoints for better terrain mapping efficiency using an UAV. To create a map of a desired terrain, an UAV is used to capture images at particular GPS locations. These images are then stitched together to form a complete map of the terrain. To generate a good map using image stitching, the images are desired to have a certain percentage of overlap between them. In high windy condition, an UAV may not capture image at desired GPS location, which in turn interferes with the desired percentage of overlap between images; both frontal and sideways; thus causing discrepancies while stitching the images together. The information about the exact GPS locations at which the images are captured can be found on the flight logs that are stored in the Ground Control Station and the Auto pilot board. The objective is to look at the flight logs, predict the waypoints at which the UAV might have swayed from the desired flight path. If there are locations where flight swayed from intended path, the code should generate a new set of waypoints for a correction flight. This will save the time required for stitching the images together, thus making the whole process faster and more efficient. / Dissertation/Thesis / M.S. Mechanical Engineering 2014

<p> Walking robots consume considerable amounts of power, which leads to short mission times. Many of the tasks that require the use of walking robots, rather than wheeled, often require extended periods of time between the possibility of charging. Therefore, it is extremely important that, whenever possible, the gait a walking robot uses is as efficient as possible in order to extend overall mission time. Many approaches have been used in order to optimize the gait of a hexapod robot; however, little research has been done on how enabling the leg segments of a hexapod to extend will impact the efficiency of its gait. In this thesis, a joint space model is defined that includes both rotational joints as well as prismatic joints for expanding and contracting individual leg segments. A genetic algorithm (GA) is used to optimize the efficiency of a gait using the joint space based on a tripod gait. Other considerations for the gait include stability and dragging, which affects overall efficiency of a gait. The results of preliminary runs of the GA show the impacts of changing the weights of a multi-objective function, the number of generations, the number of parents retained between generations and the mutation rate. Further experiments show the impact of dynamic leg lengths on the overall efficiency of a hexapod tripod gait.</p><p>

<p> A better understanding of human hand functionality can help improve robotic and prosthetic hand capability, as well as having benefits for rehabilitation or device design. While the human hand has been studied extensively in various fields, fewer existing works study the human hand within frameworks which can be easily applied to robotic applications, or attempt to quantify complex human hand functionality in real-world environments or with tasks approaching real-world complexity. This dissertation presents a study of human hand functionality from the multiple angles of high level classification methods, whole-hand grasp usage, and precision manipulation, where a small object is repositioned in the fingertips.</p><p> Our manipulation classification work presents a motion-centric scheme which can be applied to any human or hand-based robotic manipulation task. Most previous classifications are domain specific and cannot easily be applied to both robotic and human tasks, or can only be applied to a certain subset of manipulation tasks. We present a number of criteria which can be used to describe manipulation tasks and understand differences in the hand functionality used. These criteria are then applied to a number of real world example tasks, including a description of how the classification state can change over time during a dynamic manipulation task.</p><p> Next, our study of real-world grasping contributes to an understanding of whole-hand usage. Using head mounted camera video from two housekeepers and two machinists, we analyze the grasps used in their natural work environments. By tagging both grasp state and objects involved, we can measure the prevalence of each grasp and also understand how the grasp is typically used. We then use the grasp-object relationships to select small sets of versatile grasps which can still handle a wide variety of objects, which are promising candidates for implementation in robotic or prosthetic manipulators. </p><p> Following the discussion of overall hand shapes, we then present a study of precision manipulation, or how people reposition small objects in the fingertips. Little prior work was found which experimentally measures human capabilities with a full multi-finger precision manipulation task. Our work reports the size and shape for the precision manipulation workspace, and finds that the overall workspace is small, but also has a certain axis along which more object movement is possible. We then show the effect of object size and the number of fingers used on the resulting workspace volume &ndash; an ideal object size range is determined, and it is shown that adding additional fingers will reduce workspace volume, likely due to the additional kinematic constraints. Using similar methods to our main precision manipulation investigation, but with a spherical object rolled in the fingertips, we also report the overall fingertip surface usage for two- and three-fingered manipulation, and show a shift in typical fingertip area used between the two and three finger cases. </p><p> The experimental precision manipulation data is then used to refine the design of an anthropomorphic precision manipulator. The human precision manipulation workspace is used to select suitable spring ratios for the robotic fingers, and the resulting hand is shown to achieve about half of the average human workspace, despite using only three actuators.</p><p> Overall, we investigate multiple aspects of human hand function, as well as constructing a new framework for analyzing human and robotic manipulation. This work contributes to an improved understanding of human grasp usage in real-world environments, as well as human precision manipulation workspace. We provide a demonstration of how some of the studied aspects of human hand function can be applied to anthropomorphic manipulator design, but we anticipate that the results will also be of interest in other fields, such as by helping to design devices matched to hand capabilities and typical usage, or providing inspiration for future methods to rehabilitate hand function.</p>

<p> Advancements in the field of machine learning has made a model-free approach for nonlinear control of dynamical systems more viable. Traditionally, the controller design is based on the analysis of the system model. In practice, however, it might not be possible to estimate a system model that truly reflects the complex behavior of the real system. A model-free controller self-learns the required control decisions by applying machine learning techniques, avoiding the need for estimation and analytical design. In this thesis, a parameterized dynamical system known as Dynamic Movement Primitive (DMP) is used as a feedforward model-free controller. An advanced, nature-inspired, evolutionary machine learning algorithm called Covariance Matrix Adaption Evolution Strategy (CMA-ES) was used to self-learn the control decisions. It was demonstrated through computer simulated experiments that such an evolutionary model-free controller could successfully learn to accomplish the difficult task of swinging up a double inverted pendulum, motivating further research.</p><p>

An arithmetic processor chip for use in robotics has been designed in 4µm CMOS technology. The chip is intended to function as a slave processor to a general purpose microprocessor host and be able to perform robot arm coordinate transformation calculations for use in real-time control applications. A parallel-processing, multi-pipelined architecture has been used to produce a 45mm² chip for which a machine cycle time of 200ns appears possible. The general nature of the architecture of this microprogrammable processor renders it useful for a range of computational tasks in robotics in addition to coordinate transformation. / Applied Science, Faculty of / Electrical and Computer Engineering, Department of / Graduate