Jenelle Armstrong Piepmeier

Uncalibrated dynamic visual servoing

A dynamic quasi-Newton method for uncalibrated, vision-guided robotic tracking control with fixed imaging is developed and demonstrated. This method does not require calibrated kinematic and camera models. Robotic control is achieved at each step through minimizing a nonlinear objective function, by taking quasi-Newton steps and estimating the composite Jacobian at each step. The Jacobian is estimated using a dynamic recursive least-squares algorithm. Experimental results demonstrate the validity of this approach.

Uncalibrated Eye-in-Hand Visual Servoing

In this paper we present new uncalibrated control schemes for vision-guided robotic tracking of a moving target using a moving camera. These control methods are applied to an uncalibrated robotic system with eye-in-hand visual feedback. Without a priori knowledge of the robots kinematic model or camera calibration, the system is able to track a moving object through a variety of motions and maintain the objects image features in a desired position in the image plane. These control schemes estimate the system Jacobian as well as changes in target features due to target motion. Four novel strategies are simulated and a variety of parameters are investigated with respect to performance. Simulation results suggest that a Gauss-Newton method utilizing a partitioned Broydens method for model estimation provides the best steady-state tracking behavior.

Modern robotics engineering instruction

We discuss three basic principles of modern robotics education and contrast the traditional teaching style used in a majority of engineering classes with that necessary for a rapidly developing field such as robotics. Our basic tenet is that a modern robotic engineer must have knowledge, experience, and insight. While traditional education methods focus on knowledge and experience through the standard lecture-laboratory cycle, we submit that insight is the key to a complete robotics education. We conjecture that insight cannot be gained from textbooks, lectures, and laboratory exercises alone, as these tend to focus on merely academic rather than global and social issues. We present a discussion of the techniques used at the United States Naval Academy to inform, educate, and motivate students in the field of robotics.

Tracking a moving target with model independent visual servoing: a predictive estimation approach

Target tracking by model independent visual servo control is achieved by augmenting quasi-Newton trust region control with target prediction. Model independent visual servo control is defined using visual feedback to control the robot without precise kinematic and camera models. While a majority of the research assumes a known robot and camera model, there is a paucity of literature addressing model independent control. In addition, most researches have focused primarily on static targets. The work presented here demonstrates the use of predictive filters to improve the performance of the control algorithm for linear and circular target motions. The results show a performance of the same order of magnitude as compared to some model based visual servo control research. Certain limitations to the algorithm are also discussed.

The RoboCup Nanogram League: An Opportunity for Problem-Based Undergraduate Education in Microsystems

A problem-based learning approach was chosen for a new senior elective in microsystems. The problem posed to the students was to design microrobots suitable for the new ldquonanogram leaguerdquo of the international RoboCup competition, which challenges teams of students and researchers to construct microscopic untethered robots that will compete against each other in soccer-related agility drills on a 2.5 mm by 2.5 mm playing field. The approach was shown to increase student interest and motivation. The course was considered a success and will be repeated with some modifications to increase the breadth of the course coverage.

Scale-model vehicle analysis for the design of a steering controller

A scale-model vehicle is developed that is dynamically similar to a full-size vehicle through application of the Buckingham-Pi theorem. Specifically, the vehicle is modified to match the corresponding Pi groups of the scale-model vehicle and the average of a number of full-size vehicles. The modifications require the mass of the vehicle and its distribution to be changed. Furthermore, an experimental apparatus is designed and built to measure the cornering stiffness of scale-model tires. Experimental data for the vehicle is compared to data for a number of full-size vehicles.

Experimental results for uncalibrated eye-in-hand visual servoing

This paper investigates uncalibrated control schemes for vision-guided robotic tracking of a moving target using a moving camera. These control methods are applied to an uncalibrated robotic system with eye-in-hand visual feedback. Without a priori knowledge of the robots kinematic model or camera calibration, the system is able to track a moving object through a variety of motions and maintain the objects image features in a desired position in the image plane. These control schemes estimate the system Jacobian as well as changes in target features due to target motion. Four novel strategies are tested.

Steering controller design using scale-model vehicles

A scale-model, experimental apparatus is developed to design and evaluate steering controller designs. The controller is designed using linear quadratic optimal control methods on a standard four-state vehicle model. The apparatus simulates the human driver using a scale-model vehicle, a treadmill, and a vision system.

Uncalibrated target tracking with obstacle avoidance

Target tracking and obstacle avoidance are demonstrated for uncalibrated visual servoing. An objective function is designed that encourages target following by a robotic end-effector while discouraging movements near an obstacle. The objective function incorporates the error between the target and the end-effector and a potential function related to the obstacle. This objective function is minimized using a dynamic nonlinear least squares optimization method in conjunction with a recursive least squares Jacobian estimation algorithm. The approach is generic and can be applied to a variety of systems. Calibration is unnecessary after a reconfiguration or disturbance to the robotic workcell. This type of control has the potential to provide a low-cost, low-maintenance automation solution for unstructured industries and environments. Experimental results demonstrate both target tracking and obstacle avoidance for an uncalibrated robotic system.

MEMS Kinematics by Super-Resolution Fluorescence Microscopy

Super-resolution fluorescence microscopy is used for the first time to study the nanoscale kinematics of a MEMS device in motion across a surface. A device under test is labeled with fluorescent nanoparticles that form a microscale constellation of near-ideal point sources of light. The constellation is imaged by widefield epifluorescence microscopy, and the image of each nanoparticle is fit to a Gaussian distribution to calculate its position. Translations and rotations of the device are measured by computing the rigid transform that best maps the constellation from one image to the next. This technique is used to measure the stepwise motion of a scratch drive actuator across each of 500 duty cycles with 0.13-nm localization precision, 1.85-nm displacement uncertainty, and 100-μrad orientation uncertainty for a constellation diameter of 15 μm. This novel measurement reveals acute aperiodic variations in the step size of the actuator, which have been neither previously observed nor predicted by any of the published models of the operation of the device. These unexpected results highlight the importance of super-resolution fluorescence microscopy to the measurement of MEMS kinematics, which will have broad impact in fundamental investigations of surface forces, wear, and tribology in MEMS and related applications.