Camera Based Local Positional Vision-Guidance System for Micro Unmanned Ariel Vehicles

Adviser: Dr. Camillo J. Taylor

Abstract: Micro-Unmanned Aerial Vehicles (mUAVs) cannot navigate terrains with proper positional system sensors without having to sacrifice overall flight capabilities. To relieve the burden of heavy sensors on the mUAV, a camera-based vision-guided local positioning system was created. This system used Lucas-Kanade feature detection to identify the areas of interest. We implemented affine transformation properties with Gaussian-Jordan elimination to produce coordinate data on the camera’s position. From this datum and calculations, we concluded where the camera, and subsequently the mUAVs are in space locally. View Slides

PR2 Head and Hand Manipulation through Teleoperation: Using an Attitude and Heading Reference System

Advisor: Dr. Camillo J. Taylor

Abstract: Previous research using Vicon, a motion capture system, has been used to tale­ operate the PR2 robot’s head using tracking markers attached to the user’s head mounted display (HMO). However, the Vicon system requires special cameras, rigging, and an extra software suite to perform this task. Our current goal is to find a smaller, mobile alternative that is at least as accurate as the Vicon system in order to manipulate the PR2 robot, from Willow Garage. The CHR-6dm Attitude and Heading Reference System (AHRS), from CH Robotics, meets our goal. The CHR-6dm is the size of a quarter, can be easily attached to the user’s HMO, and its driver and programs can be added to the existing robots operating system. Shown to be an acceptable replacement to the Vicon system, the AHRS, in conjunction with actuators and tactile sensors, is being considered for controlling the PR2’s grippers and contributing to a more complete tele-immersion experience. View Slides

iRobot Create Navigation with Mapping Interpretation Explored Through Smart Camera Networks

Advisor: Dr. Camillo J. Taylor

Abstract: This paper describes our approach of tackling the most common problem when dealing with robotics, localization, the ability for a robot to know its current position relative to its environment and any of its previous positions. The idea is to create a system that would be used to automatically localize both the cameras and the robot relative to any characterized reference points. This project is built off of past work that used optical signaling techniques to localize a set of smart camera modules. The iRobot Create’s odometry will be used to measure precise positions of the robot in relation to the positions of the smart camera network. The smart camera modules have the ability to localize itself and send back their positions relative to one another. Once their positions and distances from one another are known it is easier to find the position of the robot relative to the positions of the smart camera modules. View Slides

Novel Proprioceptive Sensors for a Legged Robot

Advisors: Drs. Aaron Johnson and Daniel Koditschek

Abstract: We present work towards a two-part, per-leg state estimation framework on a hexapedal robot which separately measures motor core temperatures and estimates each leg’s angular offset from its gravity vector. Motor core temperature measurements are made by finding the resistance of each motor and using a simple model which calculates temperature using this resistance and known constants. Leg angle estimates are made by constructing an observer for each motor-leg combination without gravity compensation and using discrepancies between the observer and actual leg to discern the direction of gravity’s pull. The viability of these methods is demonstrated by presenting preliminary data comparing these estimated quantities with known ground truths. We believe the eventual implementation of this framework on a robot will enable highly dynamic behaviors by offering more finely-tuned thermal protection than is presently available and will reduce system complexity and fragility by supplanting the presence of an IMU where it is used for only pitch measurements. View Slides

Characterizing the Mechanical Response of the Extra Fibrillar Matrix of the Annulus
Fibrosus

Advisor: Dr. Dawn M. Elliott

Abstract: A major cause of back pain and spinal cord deformation is intervertebral disk degeneration. A complex network of tissues gives the disk anisotropic and inhomogeneous mechanical properties allowing motion, load transfer and energy absorption. These complexities make it difficult to accurately simulate the mechanical functions of the annulus fibrosus (AF), the fibrous ring of the intervertebral disc. However, computer models allow us to simulate its behavior in addition to minimizing cost and time of running experiments of scenarios that may be too complicated or dangerous to do in vivo. This study gains insight on the mechanical behavior of the AF in human samples through regional biaxial and uniaxial loading, and confined compression tests. With these data, we aim to determine which simulation model best represents the AF in motion segment algorithmic simulation. Our findings show that mechanical properties vary throughout different regions in the AF, a component which is not accounted for in current finite element models. These preliminary results will be used to develop a model that accounts for regional variations of mechanical behavior in the AF. View Slides

Mapping Magnetic Field Topography in Microrobotic Control

Advisor: Dr. Vijay Kumar

Abstract: Microrobots on the scale of 30 J.lm3 are used to organize and provide drug delivery to microorganisms in vitro. This scale limits the control of these microrobots; methods such as tethering and automation are not ideal with existing technology. Magnetic fields provide an external, non-contact force that allows for selective, precise control. In this study a magnetic field caused by four coils with ferrite cores was mapped in a 1cm2 workspace. An experimentally based model was then developed to provide the field and gradient vectors everywhere within the workspace. The force on the microrobots is directly proportional to the field gradient. In future work the motion of the microrobots will be predicted in a given field topography defined by the model, and the desired field topography for a defined path will be found when the model is inverted. View Slides

Electropolymerization of Aniline on Platinum Electrodes used for the Construction of an Electrochemical Supercapacitor

Advisors: Drs. Jorge Santiago-Aviles and Rocio del A. Cardona

Abstract: Supercapacitors have advantages over normal batteries in that they are able to discharge and charge more rapidly. Research on Faradaic devices has been geared to making supercapacitors even more energy efficient by using polymers like poly-aniline to maximize charge and storage methods. This paper presents initial result of test run with the electroactive monomer aniline with tetrabutylammoniumperchlorat (TBAP) and acetonitrile as solvent. The electrochemical properties of poly-aniline single electrodes and devices were investigated using cyclic voltammetry methods and compared with previously utilized thiophenes voltammetry data and methods. View Slides

Designing a Stress/ Strain Apparatus for Organic Field Effect Transistors

Advisor: Dr. Cherie Kagan

Abstract: In order to devise a new method to sense the activity of brain neurons, my group is currently in the process of fabricating a series of organic field-effect transistors (OFETs) on plastics. The three-dimensional nature of the brain and its lack of rigidity, require electronics that are flexible and stretchable. We will devise a device to test FETs under deformation. For this project, the forms of deformation that we expect the transistors to incur are uniaxial loading and bending. In designing an apparatus, the following parameters must be met:

• The device must be able to bend and deflect the substrate
• The device must be able to pull and compress the substrate
• The device must allow the user to probe the FET’s electrical characteristics under stress or bending.

The success of the apparatus will be based upon its ability to fulfill the criteria necessary to test the performance of the transistors in deformation. Upon completion it will give us an idea of how the transistors will behave once placed on the surface of brain. View Slides

A 3-D Electrophysiological Heart Model for Arrhythmia Simulation and Visualization

Advisor: Professor Rahul Mangharam

Abstract: Heart disease is the leading cause of death in the United States, accounting for 25.4% of the total deaths. Clinical trials in 2002 confirmed that anti-arrhythmic drugs used by 1.5 million Americans do not offer health benefits and actually increase the risk of complications. The alternative, catheter ablation surgical procedures, only yield success rates of 40-85%, thus requiring repeated procedures in over half of the cases. Consequently, there is a need of extensive cardiac electrophysiological (EP) studies to better address these medical conditions to discover the best ablation treatments. Currently, UPenn has developed a real-time Virtual Heart Model (VHM) that reflects the electrophysiological functioning of both the normal and dysfunctional heart. However, the current 2-D VHM lacks enough spatial information to assist in EP studies. Heart anatomy cannot be represented in 2-D, and therefore 3-D modeling is imperative for EP studies. To this end, a realistic 3-D heart model has been developed to (a) present more accurate anatomical spatial information, (b) study the mechanisms of arrhythmias, and (c) become a guidance tool for EP studies and ablation therapy. Such a model provides a Spatia-temporal model that captures the electrophysiological aspect of the heart tissue. It will also be a guidance tool for EP studies by displaying cardiac timing-anomalies as a heat-map around the 3-D model in order to quickly and accurately locate “timing hotspots” that cause cardiac arrhythmias. View Slides

A Piezoelectric Approach to the Mobile Monitoring of Neonatal Breathing

Advisors: Drs. Jay Zemel and Barbara Medoff-Cooper

Abstract: Neonatal feeding, characterized by a breathing, sucking, and swallowing pattern, has been shown to be of high prognostic value to neonatologists in predicting future developmental and behavioral problems in infants [1]. The bulkiness of previous neonatal monitoring has limited their usefulness. To better capture neonatal feeding patterns, a mobile device (NeoNur) has been developed that successfully monitors sucking characteristics in infants. Sucking behavior alone, however, does not offer a complete description of the neonatal feeding process. Specifically, the relation between sucking behavior and breathing behavior is of particular importance. To this end, an attachable breathing sensor was developed to gather breathing data and sucking data simultaneously. This breathing sensor, composed of piezoelectric film overlaid horizontally across a ridged ring, provides a suitable solution for capturing critical features of human respiration, including breath frequency and amplitude, while an infant feeds. These results suggest that the NeoNur equipped with the breathing sensor is a viable tool for neonatologists in collecting neonatal breathing behavior simultaneously with sucking behavior. View Slides

Pediatric In-Shoe Physical Activity Dynanometer

Advisor: Dr. Jay N. Zemel

Abstract: Osteoporosis is a prevalent disease that is responsible for over 1.5 million bone fractures each year. According to the Mechanostat model of bone growth, bone size and mass are impacted by the forces developed in muscles during growth; therefore, a device to measure forces in the legs during growth is desired. The goal of the Foot PAD project is to address the main problems currently facing such force measurement devices: obtrusiveness will be eliminated by designing a sensor circuit that is completely concealed inside of a shoe, and inaccuracy in measurements will be minimized. Piezoresistive sensors were chosen due to small size and ability to reproduce results well in changing temperatures. A circuit was designed for the device and the printed circuit board was machined in the lab. The board was then placed in a shoe with sensors attached and data were taken. Calibration curves created from data yielded linear fits in each case, which was expected from calculations, with R squared values of over 0.95 in most cases. The strong linear correlation between the actual and measured forces suggest that there will be little inaccuracy in measurements, and, as none of the users to this point have reported any discomfort or restrictions while using the device, the problem of obtrusiveness seems to have been remedied. View Slides

Specializing Single Walled Carbon Nanotunes for the Detection of Prostate Cancer Biomarkers

Advisor: Dr. A.T Johnson

Abstract: Prostate cancer is the second leading cause of cancer-related deaths in American men. While there are known treatments for prostate cancer, the later the cancer is detected, the lower the probability a treatment will be successful. Prostate cancer often goes undiagnosed until later stages because current diagnostic methods require a high concentration of cancer biomarker proteins to ensure an accurate diagnosis. Here we present field effect transistors composed of functionalized single-walled carbon nanotubes as a method for earlier and more reliable detection. Each carbon nanotube transistor possesses unique electrical properties. Our sensors’ utility is based on the fact that these properties are affected in a predictable manner by changes to the nanotubes’ functional attachments. By functionalizing the semiconducting carbon nanotubes with antibodies that bind in vivo to osteopontin-a prostate cancer biomarker-we have created a device that would produce a characteristic change in the transistors’ on-currents, turn-off voltages and mobilities when exposed to varying concentrations of osteopontin. These devices were able to detect osteopontin at concentrations as low as 1 pg/mL, which is one thousand times lower than the current ELISA detectin standard of roughly 1.0 ng/mL. View Slides

Altered Mechanosensitivity with Modulation of Nuclear Mechanics in Fibrochondrogenic Mesenchymal Stem Cells

Advisor: Dr. Robert L. Mauck

Abstract: Mesenchymal stem cells (MSCs) are of interest for tissue engineering because of their multipotent nature and regenerative applications. To better integrate the MSCs into a tissue engineered scaffold and further the differentiation process, attempts are being made to understand the underlying mechanisms of mechanotransduction in these cells. This study examined the effects of changing nuclear stiffness as a result of MSC fibrochondrogenesis on the regulation of gene expression with mechanical loading. In this work, nuclear condensation (i.e., heterochromatin levels) was decreased via the addition of the histone deacetylase inhibitor, Trichostatin A (TCA). Heterochromatin intensity, nuclear aspect ratio (NAR) with stretch, nuclear stiffness by atomic force microscopy, and cartilage­ specific mRNA levels were measured with and without TCA addition, and under the influence of static stretch of the scaffold. Consistent with our past studies, the nuclear stiffness of fibrochondrogenic MSCs increased, as did expression of the cartilage markers AGG and COL II. However, the addition of TCA significantly decreased nuclear stiffness (and increased deformation of the nucleus when scaffolds were exposed to 10% static stretch) and heterochromatin staining. Interestingly, TCA also blocked the mechanical sensitivity of MSCs to static stretch. These results show that altering the nuclear mechanics of a differentiating MSC influences the mechanosensitivity of these cells, and helps to explicate the complex and time dependent features of this differentiation process. View Slides