הפקולטה להנדסה ע"ש איבי ואלדר פליישמן

סמינר מחלקתי Avishai Beck and Roi Peretz

02 במאי 2016, 15:00

וולפסון 206

סמינר מחלקתי Avishai Beck and Roi Peretz

School of Mechanical Engineering Seminar
Monday, May 2, 2016 at 15:00
Wolfson Building of Mechanical Engineering, Room 206

A Jumper Glider robot
inspired by the locust

Avishai Beck

MSc Student of Dr. Gábor Kósa

In recent decades, researchers began to look for new solutions for mobility of robots by imitating the biological mechanisms found in nature, especially insects’ locomotion ability. Our research group engaged in developing a miniature jumping and gliding robot that mimics the locust’s locomotion. The target of the robot is to be able to maneuver and navigate in rough terrain by producing continues controlled jumps.

The goal of this study was to develop the gliding mechanism and then integrate it with the TAUB jumping mechanism in order to build a jumping gliding robot capable of multiple jumps and changing its orientation in air.

The robot jumps and commences a ballistic trajectory with closed wings and tail to reduce its drag. When the robot reaches the apex of the jump it spreads open the wing and tail and then glides down in a controlled descent. The gliding phase has three main objectives: To enhance the achieved distance, reduce the velocity of the landing impact from the ground and to determine the yaw angle of the body at landing.

In order to characterize the robot, a dynamic model with six degrees of freedom was developed. A model is solved numerically using the Matlab platform. The model aided in the analysis of the robots stability during in the airborne phase of its jump and the assessment of the gliding and yaw control abilities.

Sebveral prototypes were designed and evaluated experimentally to examine the separate phases of the robot’s jump. We tested the wings opening mechanism, stability of the ascent, gliding performance and yaw controlling. After developing the final gliding mechanism it was combined with the TAUB jumping mechanism and experiments were taken out on the fully integrated robot.

The latest prototype of the Jumper-Glider robot weighs 28 [gr]. It is capable of jumping to a height of 1.7 [m] and a distance of 4 [m]. The raobot controls its yaw angle in the air and it is capable of continuous jumps.

School of Mechanical Engineering Seminar
Monday, May 2, 2016 at 15:00
Wolfson Building of Mechanical Engineering, Room 206

Ozonation of lignocellulosic biomass as a pre-treatment step

for bioethanol production

Roi Peretz

MSc Student of Assoc. Prof. Hadas Mamane and Dr. Yoram Gerchman (Haifa University)

Geo-political concerns, environmental and health issues, and the growing need for energy have increased the demand for developing a new sustainable and clean source of energy for transportation, heating and industrial processes. There is a great need for “drop-in” fuels which will directly replace the petroleum based fuels used today. Unlike other energy products (like heat and power) which can be replaced by most of the alternative solutions, for the transportation sector there seems to be no alternative solution other than biomass conversion to energy. The goal of this research was to examine the use of ozone based Advanced Oxidation Processes (AOP’s) as a pre-treatment method of lignocellulosic biomass (such as agricultural waste) prior to cellulase treatment used for conversion of cellulose to sugars. The research aimed to use ozonation on highly concentrated samples, as part of a vision for promoting a new, economic and competitive process. The ozonation experiments were first conducted on phenol (lignin) models such Gallic Acid and Tannic Acid with concentration up to 60 gr/L. Triple phase kinetics with two transition points have been obtained. Enzymatic activity demonstrated the highest potential at these two transition points. These results interestingly suggest that full mineralization of the phenols is not necessary for an optimal cellulase enzymatic activity. On later stages ozonation was conducted on biomass material such as olive mill solid wastes (OMSW) and paper wastes. It was found that post-ozonation samples of paper have lost their orderly fiber-type structure, due to the diminishing of the lignin part of the mass. In both cases, full decomposition of total phenols was achieved in a relative short periods of times. The combination of the above could make lignocellulosic biomass saccharification and fermentation processes more feasible, thereby rendering the conversion of agricultural waste more realistic. Moreover, creation of a new and ubiquitous source of biofuels is unequivocally a major goal of the biofuels research community.

03 דצמבר 2015

ידיעות נוספות בנושא

EE Seminar: Improving parallel programs with architectural insights

~~(The talk will be given in English)

Speaker: Dr. Adam Morrison
CS, Technion

Wednesday, December 23rd, 2015
15:00 - 16:00
Room 011, Kitot Bldg., Faculty of Engineering
Improving parallel programs with architectural insights
Parallel programs and concurrent algorithms are developed using models---formal and mental---that abstract away details of the underlying hardware. In practice, however, these hidden details can have a dominating impact on performance. In this talk, I will show
how understanding hardware/software interactions can lead to more efficient programs and algorithms, through improvements of both models and hardware:
(1) I will show how understanding the implementation of the TSO memory model by x86 processors enables designing a fence-free work stealing algorithm, which was thought to be impossible.
(2) I will describe how insights on the way software uses atomic memory operations inspire a novel hardware architecture design that can break the serialization of certain contending synchronizing atomic instructions, and execute them in parallel.

23 בדצמבר 2015, 15:00

חדר 011, בניין כיתות-חשמל

EE Seminar: From representation to inference: respecting and exploiting mathematical structures in computer vision and machine learning

~~(The talk will be given in English)

Speaker: Dr. Oren Freifeld
CS and Artificial Intelligence Lab, MIT

Monday, January 11th, 2016
15:00 - 16:00
Room 011, Kitot Bldg., Faculty of Engineering
From representation to inference: respecting and exploiting mathematical structures in computer vision and machine learning

Abstract
Stochastic analysis of real-world signals consists of 3 main parts: mathematical representation; probabilistic modeling; statistical inference. For it to be effective, we need mathematically-principled and practical computational tools that take into consideration not only each of these components by itself but also their interplay. This is especially true for a large class of computer-vision and machine-learning problems that involve certain mathematical structures; the latter may be a property of the data or encoded in the representation/model to ensure mathematically-desired properties and computational tractability. For concreteness, this talk will center on structures that are geometric, hierarchical, or topological.

Structures present challenges. For example, on nonlinear spaces, most statistical tools are not directly applicable, and, moreover, computations can be expensive. As another example, in mixture models, topological constraints break statistical independence. Once we overcome the difficulties, however, structures offer many benefits. For example, respecting and exploiting the structure of Riemannian manifolds and/or Lie groups yield better probabilistic models that also support consistent synthesis. The latter is crucial for the employment of analysis-by-synthesis inference methods used within, e.g., a generative Bayesian framework. Likewise, imposing a certain structure on velocity fields yields highly-expressive diffeomorphisms that are also simple and computationally tractable; particularly, this facilitates MCMC inference, traditionally viewed as too expensive in this context.

Time permitting, throughout the talk I will also briefly touch upon related applications such as statistical shape models, transfer learning on manifolds, image warping/registration, time warping, superpixels, 3D-scene analysis, nonparametric Bayesian clustering of spherical data, multi-metric learning, and new machine-learning applications of diffeomorphisms. Lastly, we also applied the (largely model-based) ideas above to propose the first learned data augmentation scheme; as it turns out, when compared with the state-of-the-art schemes, this improves the performance of classifiers of the deep-net variety.

Speaker's bio:
Oren Freifeld is a postdoc at John Fisher's Sensing, Learning and Inference group at MIT CSAIL. His main research areas are computer vision, probabilistic modeling, statistical inference, and machine learning. He is also interested in stochastic processes, image processing, statistical signal processing, and large-scale data analysis. He earned his PhD (advisor: Michael Black) and ScM in Applied Mathematics from Brown University. During parts of his PhD he was a Visiting Scholar at Stanford University Electrical Engineering department (host: Krishna Shenoy). Earlier, he graduated from Tel-Aviv University with MSc (advisors: Hayit Greenspan and Jacob Goldberger) and BSc in Biomedical Engineering.

11 בינואר 2016, 15:00

חדר 011, בניין כיתות-חשמל

סמינר מחלקתי Dr. Steve Levine

16 בדצמבר 2015, 15:00

וולפסון 206

School of Mechanical Engineering Seminar
Wednesday, December 16, 2015 at 15:00
Wolfson Building of Mechanical Engineering, Room 206

Breaking the Barriers to Realistic Simulation of the Human Body

Dr. Steve Levine

Living Heart Project at Dassault Systems, SIMULIA

It would be difficult to overestimate the impact on engineering, industry and society if we can accurately and cost effectively perform realistic simulations of the human body and its systems. In fact, understanding and replicating the physics of the human body seems poised to be the next major horizon for a growing range of simulation tools that have now become irreplaceable for developing products from microchips to jetliners. However, despite being the focus of countless studies over decades, it is not until recently has instrumentation been available to reveal its detailed dynamic behavior with sufficient accuracy to allow modeling and simulation to finally cross the threshold onto this frontier.

This presentation will highlight several areas of significant progress in virtual human simulation including the global translational cardiovascular initiative, the Living Heart Project. Through this project, a quantitatively accurate, fully coupled electromechanical simulation of a four chamber human heart was developed and made generally available in a robust commercial package. This computational model can provide critical insight to improve and develop new devices and procedures for treating heart conditions, the leading cause of death worldwide. Progress on musculoskeletal hybrid multiscale models will also be presented that consist of realistic 3D structures for deformable joints, ligaments and soft tissues using novel non-linear finite element workflows to isolate the force balance under true load conditions.

Applications examples will demonstrate the state-of-the art in replicating anatomically correct behavior at both the organ and musculoskeletal level using standard technology, representing a scalable translational environment for understanding human behavior in a real world context.

Bio:

Dr. Steve Levine is Sr. Director of Product Strategy and the Executive Director of the Living Heart Project at Dassault Systems SIMULIA. Since 2006, Steve has been responsible for driving the SIMULIA strategy towards its vision of enabling simulation to help harmonize Product, Nature and Life. Steve began his career at Engelhard Corporation, where he founded and led an applied simulation group in corporate R&D for over five years. In that time, his work featured in a number of discoveries, new product patents and he received several technical awards, including the highest research achievement award possible. Steve subsequently joined Biosym Technologies, a biomolecular simulation startup, which grew to become the leading modeling provider for the Pharmaceutical, Chemical, Materials industries, going public as Accelrys Inc. in 2003. Acclerys is now part of Dassault Systems creating the foundation for the brand BIOVIA. During his 15 years at Accelrys, he held a number of positions including Director of Product Development, GM of Materials Informatics, and Sr. Director of Corporate Development. Steve holds a Ph.D. in Materials Science and Engineering from Rutgers University and has been elected into the American Institute for Medical and Biological Engineering (AIMBE)'s College of Fellows.

סמינר מחלקתי - יאיר זיק

Towards the Deployment of Cooperative Game-Theoretic Solutions

22 בדצמבר 2015, 14:00

Abstract :

Cooperative game theory spans the formation of coalitions among collaborative agents, as well as

proposing reasonable payoff divisions among them. This branch of game theory is rooted in Von-

Neumann & Morgenstern’s foundational work, with many beautiful theoretical ideas; however, it has seen

relatively sparse application. In this talk, I will discuss several research thrusts which aim at making the

theory of cooperative games more applicable; I will discuss how the introduction of overlapping coalition

structures – i.e. allowing agents to divide their resources among more than one coalition – allows one to

model complex agent interaction.

Moreover, I will show how one can overcome the computational challenges traditionally associated with

finding cooperative solution concepts by relaxing our requirements. By looking for a probably

approximately correct (PAC) solution, and applying ideas from computational learning theory, one can

find good solutions to cooperative games while eliminating computational overhead.

Finally, I will discuss exciting directions for the study of cooperative games, both in the application of the

theory to causality and classification, and in empirical human trials

Short Bio:

Yair Zick is a postdoctoral research fellow in the computer science department at Carnegie Mellon

University. He has completed his PhD at Nanyang Technological University, SPMS (funded by the

Singapore A*STAR SINGA award). He received his B.Sc (Mathematics and the "Amirim" honors program)

from the Hebrew University of Jerusalem. His research interests include game theory, fair division and

their applications to domains such as machine learning, security, and privacy.

He is the recipient of the 2014 IFAAMAS Victor Lesser Distinguished Dissertation Award, and the 2011

Pragnesh Jay Modi Best Student Paper Award.

EE Seminar: Field Classification, Modeling and Anomaly Detection in Unknown CAN Bus Networks

~~Speaker: Moti Markovitz,
M.Sc. student under the supervision of Prof. Avishai Wool

Wednesday, December 16, 2015 at 15:00
Room 011, Kitot Bldg., Faculty of Engineering

Field Classification, Modeling and Anomaly Detection in Unknown CAN Bus Networks

Abstract

A controller area network (CAN bus) is commonly used for in-vehicle communication. During the last years, several security research groups have shown that CAN bus networks are vulnerable to attacks. In particular, it was demonstrated that an attacker can broadcast fake messages on the CAN bus network and disrupt the normal operation of the vehicle’s critical systems, like shutting down the engine during trip, spinning the steering wheel, etc.

In this thesis we deal with these attacks.
First, we describe the architecture of the vehicle communication system and the security issues. Then we explain how we acquired the data for our research, the initial analysis of the data, and what we have learned from it.
A serious challenge in in-vehicle communication research is that the CAN bus message formats are proprietary and not publicly documented. We describe the field classification algorithm that we have developed, that automatically parse the messages and find their semantics.

In order to evaluate our methods we needed traces of messages whose ﬁeld structure is known. We present a simulator of CAN bus communication, that we have developed for this purpose.

Finally, we use the message field structures from our field classification algorithm, for building an enforcement model, based on TCAM. This enforcement model can be used for anomaly detection system.

16 בדצמבר 2015, 15:00

חדר 011, בניין כיתות-חשמל

שלושת הסודות ליצירת הזדמנויות עסקיות

סדנה אקטיבית המתמקדת בסיוע לחברות ללמוד, לתרגל ולשלוט

בשלושת אלמנטים קריטיים שיסייעו להן לפתח הזדמנויות עסקיות חדשות לצמיחה וההתפתחות עסקית.

הסדנה מועברת על ידי יריב לוטן וגיל טרן המייסדים של מינטרו Marketplace

בקישור חברות להזדמנויות עסקיות ברחבי העולם.

01 בדצמבר 2015, 18:30

בניין וולפסון חדר 206 קומה 2, הפקולטה להנדסה אוניברסיטת תל אביב

איך להגיע ישירות ללקוחות הפוטנציאליים?

למשקיעים הנכונים? ואל הערוצים שתמיד רצית?

3 הסודות ליצירת הזדמנויות עסקיות

ההשתתפות ללא עלות לבוגרי הפקולטה ובעלות סמלית לאורחים נדרשת הרשמה מראש

שלושת הסודות ליצירת הזדמנויות עסקיות

סדנה אקטיבית המתמקדת בסיוע לחברות ללמוד, לתרגל ולשלוט

בשלושת אלמנטים קריטיים שיסייעו להן לפתח הזדמנויות עסקיות חדשות לצמיחה וההתפתחות עסקית.

הסדנה מועברת על ידי יריב לוטן וגיל טרן

המייסדים של מינטרו Marketplace בקישור חברות להזדמנויות עסקיות ברחבי העולם.

01 בדצמבר 2015, 18:30

בניין וולפסון חדר 206 קומה 2, הפקולטה להנדסה אוניברסיטת תל אביב

איך להגיע ישירות ללקוחות הפוטנציאליים?

למשקיעים הנכונים? ואל הערוצים שתמיד רצית?

3 הסודות ליצירת הזדמנויות עסקיות

ההשתתפות ללא עלות לבוגרי הפקולטה ובעלות סמלית לאורחים נדרשת הרשמה מראש

EE Seminar: The Role of Interaction in Streaming, Economics and Parallel Computing

~~(The talk will be given in English)

Speaker: Dr. Omri Weinstein
New York University

Wednesday, December 9th, 2015
15:00 - 16:00
Room 011, Kitot Bldg., Faculty of Engineering

The Role of Interaction in Streaming, Economics and Parallel Computing
Abstract
Over the past three decades, communication complexity has been extensively used to capture the fundamental limitations in diverse areas of computer science and modern computing systems, such as distributed (e.g., "MapReduce") platforms, data streaming, economic markets and social and physical networks.
In this talk I will describe new tools and applications of communication complexity to streaming, economics and parallelization: We prove tight bounds on the fundamental problem of approximating the "frequency moments" of a data stream, one of the most important problems in real-time monitoring of network traffic. We will then see how information and communication complexity improved our understanding of the power and limitations of parallel computing. Finally, we will explore the economic problem of distributed equilibrium computation. Our result implies that any market dynamic that converges even to an approximately stable market state, requires exponential communication in the size of the market.

BIO: Omri Weinstein is a Simons Society Junior fellow, hosted by Courant Institute (NYU). He obtained his PhD from Princeton University, under the supervision of Mark Braverman, and holds a BSc in mathematics and computer science from Tel-Aviv University. His main research lies in the intersection between interactive communication and parallel computing, economics and privacy. He has done foundational work in the emerging field of Information Complexity and its applications. His awards include the Simons Society fellowship, the Simons
graduate award in Theoretical Computer Science and the Siebel scholarship.

09 בדצמבר 2015, 15:00

חדר 011, בניין כיתות-חשמל

של משרד המדע

ידיעות נוספות בנושא

עמודים