סמינר מחלקתי

10 בפברואר 2015, 14:00 
חדר 206 בניין וולפסון  

 

Cognitive Robotics on the Factory Floor

Dr. Erez Karpas – MIT

 

Abstract:

Current manufacturing robots are typically preprogrammed to perform one fixed task efficiently. This requires a long, expensive, setup in order to integrate into an assembly line, and means making changes in the assembly process is difficult as well. Cognitive robots, which are endowed with the ability to reason, allow for faster setup through high level programming, and added flexibility in the assembly process. In this talk, I will present recent work on how cognitive robots can work with a human, responding to the human's actions intelligently, and on how they can deal with novel situations by coming up with a new course of action through replanning.

 

Bio:

Erez Karpas is a post-doctoral fellow at the Model-based Embedded and Robotic Systems group at MIT's Computer Science and Artificial Intelligence Lab. His research interests include autonomous systems, automated planning, heuristic search, and machine learning. He received his Ph.D. from the Technion - Israel Institute of Technology in 2012, and his M.Sc. and B.Sc. from Ben-Gurion University.

 

 

 

ההרצאה תתקיים ביום שלישי 10/2/15, בשעה 14:00 בחדר 206, בנין וולפסון הנדסה, הפקולטה להנדסה, אוניברסיטת תל-אביב

צוות סטודנטים מהנדסה מכנית, אונ' ת"א – לתחרות בניית והטסת מזל"טים

01 פברואר 2015

צוות סטודנטים מהנדסה מכנית,  אונ' ת"א – לתחרות בניית והטסת מזל"טים

DBF (Design Build and Fly) היא תחרות סטודנטים בין לאומית המתקיימת מדי שנה בארה"ב המאורגנת ע"י ה-AIAA (American Institute of Aeronautics and Astronautics), ובחסות חברת Cessna ו- Raytheon Missile System. הסטודנטים נדרשים לתכן, לבנות ולהטיס כלי טיס המונע חשמלית ונשלט ע"י רדיו, כאשר כל שנה המשימות לביצוע משתנות ומדמות מצבי אמת בחיי היום יום (משימת קרקע ושלוש משימות אויר).

התחרות השנה מיועדת להיערך בטוסון, אריזונה ארה"ב.

זו השנה השנייה שצוות מאונ' ת"א משתתף בתחרות. הצוות מונחה ע"י מר שלמה צח ואלי כ"ץ בעלי ניסיון של עשרות שנים בתחום. כחלק ממסקנות השנה שעברה בה זכו הסטודנטים מאוניברסיטת ת"א במקום ה-15 המכובד, מתוך 80 קבוצות משתתפות, השנה הושם דגש עיקרי על הורדת המשקל, המהווה חלק משמעותי מהציון בתחרות, תוך עמידה במשימות,  לשם כך נעשו שינויים בגישת התכן הייצור והבנייה, המתבטא בשימוש נרחב יותר בחומרים מרוכבים ותבניות. בתמונות נראות התבניות לייצור גוף המטוס והגוף הבנוי.

השנה ההרשמה לתחרות הוגבלה ל-100 אוניברסיטאות מרחבי העולם, כאשר בהן עשרות משלחות מהמזרח התיכון (טורקיה, מצרים, איחוד האמירויות וכד') כ-16 אוניברסיטאות, מאסיה (פקיסטן, סין, הודו וכד') כ-17 אוניברסיטאות, מאירופה כ-7 אוניברסיטאות וארה"ב 57 אוניברסיטאות, כולל הטובות שבהן.

ישראל משתתפת בתחרות משנת 2004, כאשר בעבר השתתפו קבוצות מהטכניון, השנה זו השנה השנייה, בה ישראל תיוצג ע"י אוניברסיטת תל אביב וזאת גם כן בהנחייתו של מר שלמה צח מחברת APG . בעבר היו מספר גופים אשר תמכו במשלחת, תמיכה שאנו מתקשים לקבל השנה.

צוות הסטודנטים בראשותו של אלעד ונטורה, נלהב לייצג את ישראל ואוניברסיטת תל אביב, כאשר הוצבה מטרה מטרה להתברג בין 10 הקבוצות הראשונות.

על מנת להגשים את החלום ולהשתתף בתחרות, עדיין חסרים כמה עשרות אלפי שקלים, שיאפשרו מימון הנסיעה לתחרות והוצאות עלות ההשתתפות בחו"ל (הפקולטה להנדסה באוניברסיטת תל אביב מממנת את כל תהליך הפיתוח, הייצור והבנייה, מתקציבי הוראה מוגבלים).

ידיעות נוספות בנושא

8.2.15

08 בפברואר 2015, 15:00 
Kitot 011  
8.2.15

Physical Electronics Dept. SEMINAR

 

***** Seminar *****

You are invited to attend a lecture

by

 

Nir Atzmon

 

(Msc. student under the supervision of Prof. Eshel Ben-Jacob and Prof. Avraham Gover)

School of Electrical Engineering, Tel-Aviv University, Tel-Aviv 69978, Israel

 

 

Real-Time feedback stimulator for investigating neuroplasticity in cultured networks

 

 

In-vitro neuronal cultures offer exciting possibilities for studying neuronal information coding and processing. A network level approach is necessary to fully understand information processing in the brain, as every function in the human brain employs correlated activities of large populations of neurons. It was shown that synaptic plasticity plays a crucial role in the synchronized behavior of the network, and therefore investigating neuroplasticity in cultured networks may yield new insight as to the functions of neuronal circuits in the brain. While much is known on the cellular mechanism underlying synaptic plasticity, making the leap from the cellular level to the network level is far from trivial.  Previous studies aimed at uncovering the dynamics governing network plasticity have used open-loop assays to study the response of neuronal networks to external stimulation by combining electrical stimulation of cultured networks on multi-electrode arrays, in conjunction with extracellular recording.

In this work we developed a closed-loop assay for investigating neuroplasticity in cultured networks. The assay makes use of a real-time feedback stimulator (RTFS) that responds to network activity within a few milliseconds and is both generic and easily modified to suit different experimental setups. In order to demonstrate the potential of such a system in the study of network plasticity, we used the RTFS to generate electrical stimulation to the network based on real-time detection of spontaneous burst patterns and were able to show a decrease in inter-burst correlation, indicating attenuation of the synchronized activity of the network. These findings imply that similar closed loop assays can be used in the study of network-level plasticity, and a generic system, such as the one presented here, can open up a wide range of research possibilities, for example in the study of epilepsy. 

 

Sunday, February 8, 2015, at 15:00

Room 011, EE- KITOT building

 

 

 

EE Seminar: Guy Nadav

~~
Guy Nadav, 
M.Sc. student under the supervision of Prof. Nahum Kiryati and Dr. Dafna Ben-Bashat

Monday, February 16, 2015 at 15:00
Room 011, Kitot Bldg., Faculty of Engineering

Pharmacokinetics Parameters Estimation from MRI

Abstract

We consider the problem of estimating the underlying anatomical structure and dynamics in brain pathologies from a set of 3-D images of the brain following the injection of a contrast agent, acquired using a Magnetic Resonance Imaging (MRI) method. The goal is to produce quantitative clinically relevant maps of the brain that can help in the assessment of neurological pathologies such as disruption of the Blood-Brain-Barrier (BBB) in a tumor, decrease of blood flow in a stroke etc.
To do that, a contrast agent (Gadolinium) is injected to a subject and temporal tracing of its concentration in the brain is conducted. Dynamic contrast-enhanced MRI (DCE-MRI) is a functional MRI method that enables temporal tracing, where  -weighted MR images are acquired dynamically after bolus injection of a contrast agent. The brain tissue effect on the measured contrast agent concentration over time (concentration-time-curve), is modelled as an Impulse Response Function (IRF). The IRF is parameterized by Pharmacokinetics (PK) parameters that represent physiological characteristics, and explain the change in the concentration-time-curve from artery to tissue voxel.
This work describes the entire process of deriving pharmacokinetics parameters from an MRI signal, and suggests novel techniques for handling estimation issues originating from the problem of accurately measuring concentration-time-curve in every voxel of the brain. We first present a few possible brain models, then describe the work that has been previously done and continue by suggesting improvements to the pharmacokinetics parameters estimation. The estimation process was tested in simulation and on data acquired from patients.
The proposed optimization method, denoted ACoPeD (AIF-Corrected-Perfusion-DCE-MRI), showed significant correlation (r=0.46, p<0.001) between flow parameter extracted from DCE-MRI and DSC-MRI (a commonly used perfusion MRI method) in patients, and is applicable in clinical settings.
In conclusion, this study proposes an optimized method, ACoPeD for tissue perfusion and permeability estimation using DCE-MRI, which is applicable in clinical settings, and includes recommendations for DCE-MRI data acquisition and analysis.

16 בפברואר 2015, 15:00 
חדר 011, בניין כיתות-חשמל  

EE Seminar: Mor Cohen

~~
Mor Cohen, 
M.Sc. student under the supervision of Prof. Yoel Shkonisky and Prof. Arie Yeredor

Monday, February 16, 2015 at 15:30
Room 011, Kitot Bldg., Faculty of Engineering

Detecting Common Lines In Cryo-EM Images Using Maximum Likelihood

Abstract

The reconstruction of three dimensional structures from Cryo-EM images is often based on the “Angular Reconstitution” method in which a coordinate system is established from three projections, and the orientation of the particle in each image is deduced from common lines shared among the images. Detecting these common lines is usually performed by first denoising the images, which takes advantage of the large number of available images to improve their SNR, followed by computing correlations between radial lines in the images. This work presents a maximum likelihood (ML) algorithm for detecting common lines that takes advantage of the highly redundant data in a different way. The algorithm is based on a global iterative detector in which we jointly estimate and classify common lines using the data from all projection images. We demonstrate by simulations that the algorithm improves the detection rate of common lines compared to state of the art methods, and operates well even with non-white imaging noise.

16 בפברואר 2015, 15:30 
חדר 011, בניין כיתות-חשמל  

Seminar 1.2.15

01 בפברואר 2015, 14:00 
Room 101, Computer and Software Engineering building  
Seminar 1.2.15

You are invited to attend a lecture

by

 

Lihi Shiloh

 

(MSc. student under the supervision of Prof. Avishay Eyal)

School of Electrical Engineering, Tel-Aviv University, Tel-Aviv 69978, Israel

 

 

Optical fiber sensing via Optical Fractional Fourier transform Reflectometry (OFFR)

 

Fiber-optic sensing is a highly promising technology for a variety of applications, ranging from biomedical sensing to structural health monitoring. In the field of Distributed Acoustic Sensing (DAS) there is a growing demand for a fiber optic sensing system that can provide fast scan rates over long distances with high spatial resolution and high sensitivity. Among the various approaches for implementing DAS, methods based on Rayleigh backscattering standout as practical and affordable solutions due to the use of a conventional communication fiber as the sensing element and the simplicity of the interrogation scheme. There are two main approaches: The first approach is Optical Time Domain Reflectometry (OTDR) in which the interrogation waveform is a short pulse and the detected light is analyzed in the time domain. The second approach is Optical Frequency Domain Reflectometry (OFDR) where the interrogation waveform is linearly chirped and the received reflections are mixed with a reference and analyzed in the frequency domain.

Recently we introduced a new interrogation method, the Optical Fractional Fourier transform Reflectometry (OFFR). It enables interrogation of distant sections of the fiber with a high scan rate and enhanced spatial resolution. It is similar to OFDR but the frequency sweep period has both positive and negative slopes. Under certain conditions the frequency variations of the backscattered light from distant reflectors and the reference light have opposite slopes. In such cases the beat signal at the receiver output will have a component whose instantaneous frequency varies linearly in time. There is a variety of methods to detect such a linearly chirped signal buried in measured data. Here we describe the use of the Fractional Fourier Transform (FrFT) to that end. We present theoretical formulation of the method and its limitations, numerical simulation of its performance in comparison with OFDR and a static, as well as dynamic, experimental results.

 

Sunday, February 1, 2015, at 14:00

Room 101, Computer and Software Engineering building

 

EE Seminar: Gilad Kahala

~~Gilad Kahala
M.Sc. student under the supervision of Dr. Hedva Spitzer and Prof. Shai Avidan

Wednesday, March 4, 2015 at 15:00
Room 011, Kitot Bldg., Faculty of Engineering

Multi Scale Blood Vessels Detection and Segmentation in Breast MRI

A method is proposed to perform segmentation of blood vessels in 3D breast MRI. The blood vessels play an essential role as an additional tool to detect tumors. Radiologists use a maximum intensity projection for the understanding of vasculature. The breast is a challenging organ in detecting vascular structures, because of noise and fat tissues. There are several existing algorithms for the detection of blood vessels in MRI images, but these usually prove insufficient when it comes to the breast. Our algorithm provided a 3-dimensional model of the blood vessels by utilizing texture enhancement followed by hessian-based methods. In addition to this, we tackled blood vessel completion by employing center line tracking, where the seeds are the end points of detached blood vessels found through skeletonizing. The results were compared to the manually segmented golden models defined by radiologist in 20 different patients, which yielded an 88.73% match to the ground truth with 13.63% false positives. It appears that with the application of mass detection as the last step, our algorithm provides a helpful tool for tumor enhancement and an automated detection of breast cancer.

04 במרץ 2015, 15:00 
חדר 011, בניין כיתות-חשמל  

Seminar 29.1.14

29 בינואר 2015, 14:00 
Kitot 011  
Seminar 29.1.14

Physical Electronics Dept. SEMINAR

 

***** Seminar *****

 

You are invited to attend a lecture by:

 

Prof. Nader Engheta

 

H. Nedwill Ramsey Professor

University of Pennsylvania

Philadelphia, PA 19104, USA

 

 

On the subject:

 

 

Metastructures for Manipulating Light-Matter Interaction

 

 

Abstract

 

In this talk, I will give an overview of some of our ongoing research activities in the areas of metamaterials, metasystems, and nanoscale optics. I will discuss some of the following topics:  (a) the extreme-parameter nanophotonics, (b) nanomaterials that perform mathematical operations (nanoscale analog computers), (c) nonreciprocal nanostructures for unusual flow of photons, (d) metamaterial “bits” and “bytes” as building blocks for digital metamaterials, and (e) graphene photonics, to name a few.  In my group we are exploring new classes of phenomena and potential applications in engineering functional metastructures.  I will present our latest results, and forecast future directions and possibilities.

 

 

Thursday, 29 January 2015 at 14:00

Room 011, Kitot Build.

 

 

29.1.15

29 בינואר 2015, 15:00 
011 Kitot  
29.1.15

Paul C. McIntyre

 

Professor and Department Chair, Materials Science and Engineering, Stanford University, Stanford, CA

 

 

On the subject:

 

 

Nanoscale Photonic and Photo-Electrochemical Behavior of Group IV Semiconductor Structures

 

This talk will summarize recent work from our group on two topics 1) photo-generated carrier dynamics in single crystal germanium nanowire arrays, and 2) the application of atomic layer deposition to protect silicon absorbers from corrosion during water splitting, while simultaneously provided good electrical connection to an overlying catalyst layer.  The first topic is related to future application of germanium nanostructures in integrated photonics.  The second has applications in renewable synthesis of fuels and chemicals.  

 

 

Thursday, 29 January 2015 at 15:00

Room 011, Kitot Build.

 

עמודים

אוניברסיטת תל אביב עושה כל מאמץ לכבד זכויות יוצרים. אם בבעלותך זכויות יוצרים בתכנים שנמצאים פה ו/או השימוש שנעשה בתכנים אלה לדעתך מפר זכויות
שנעשה בתכנים אלה לדעתך מפר זכויות נא לפנות בהקדם לכתובת שכאן >>
אוניברסיטת תל-אביב, ת.ד. 39040, תל-אביב 6997801