©2017 BY KCG-LAB


A climbing robot - Boaz Ben Moshe, Nir Shvalb, Moti shani.

Comuting radio-maps is a challenging task and requires signifcant CPU resources. In particular modern RF Propagation models take into consideration interferences, multi-paths and complex Rf-prediction
calculation such as multi-level ray tracing. We make use of Nvidia's GPGPU platform (CUDA) in order to
compute state-of-the-art RF propagation over complex terrains. This elminates the need to use large scale cluster of computers (or super computers) for this tas We assume one of the following:
(1) Urban WiFi: In-Door 3D Ray-tracing for 2.4GHz WiFi channel and signal strength predication. This work is being done as a part of the israel Short Range Consotuim.
(2) Urban and suburban WiMax: A 2.5D ray-tracing and beam tracing low frequency WiMax & LTE propagation models  and parallel Radio-map calculation implementation.
(3) GPS & Galileo Channel Simulator: implementing a multi-psth
simulator  for global position satellite  communication in urban resunding (both L1 & L2 frequencies & modulations).


An eight DOF autonmous  climabing robot is at the core of this project .
The robot is designed to climb metal trusses using magnetic grasping.
We designed  an open-loop ctuated nine-link robot robot inspired by an inchworm.
The need for fast computing of its maneuvers is obvious, Yet is us hard task considering the vast number of options available for the robot
(a traditional motion planning algorithm would result in 10 minutes of computing for each robot step).

This is your Project Description. It’s a great place to describe your Project in more detail. Add images and provide visitors with essential information about your work, including the project length, partnerships or any successful outcomes. To hook new potential clients, show how your work has provided solutions for past clients.


ViRob - A crawling micro robot - Oded Salomon, Moshe Shoham, Nir Shvalb, Yakir Nagar, Mamor Alter, boaz Ben moshe

Autonomous motion down a lumen generally involve a realtively complicated mechanism with multiple moving parts, which restricts miniaturization because of current technological limitations. Moreover those techniques often require sequential controlled actuation, thus complicating their operation. We propose an autonomous robotic device and method for crawling along a juxtapsed surface or surfaces using random vibrations. The device utilizes the effects of the anisotropic friction propeties between the surfaces along the effects of the vibrations on the anisotropic friction properties between the surfaces along the length of fibers attached  the device, and the surrounding juxtaposed suface or surfaces.The surrounding surface or surfaces may  be the unner wall of a lumen, two bounding walls for confined planar motion , a plane on which the device is disposed for two dimensional planar
motion, or an inner guide wire on which the device crawls.


Shamil Ploit, Boaz Ben Moshe, Nir Shavalb

This research project focused on an experimental device realted to virtual  environments for blind people . The idea is usse an Airbrush to "shoot air" in a modulated manner onto the hairy skin next to the wrist,  and more precisely, onto the hairs themselves, which function as cutaneous mechanical-receptors or sensors. The study, which is conducted on both blind and blindfolded  participants aims to investigate perceptual qualities of those hairs in response to modulated air stimuli. The theoretical background mostly derives from fiekds of tactile perception, thugh  the suggested approach is in fact  touch-less. The experrimntal prototypr drveloped in our lab is composed from three parts: a depth camera. (!!) a processing unit(translating the geometric surrunding to a set of air pulses).An XY robot with air pulse control.
Several types of experiments are conducted in order to examine the effectineness of the air-brush: path detection, object recognition and indoor navigation tasks. We have also designed an advance air-brush device which has the following additional eye tracking device: allowing he user to scan region without moving the head.


MeMic is a novel Anthropomorphic Hand for assisting in laparodcopic Surgery. The robot is designed to operate through a stastionary entry point  via a parallelogram linkage orientation controller as well as bending at the elbow joint. The robot is modile and small consisting of a 15 DOF anthropmophic arm and gripper as well as an intuitive glove control mechsnism. MeMic configuration allows for easier mobility, increased precision and simple to master surgerues compared with state of the devices.There are many possible surgical applications: tissue manipulation, inside instruments manipulation, vessels seizing, wound closing, electro-surgery and tissue cutting. These functions  sre useful in diverse medical fields such as: urology, gynecology, cardiothoracic
MeMic's end effector consists of three digits each having three DOF. Using three fingers instead of two allows for seceral different operational advantages. First, it allows orholding with two of the fingers and cutting it with the third. Moreover, manipulating the arc of the fingers improves the unlike many commercial robots today, which have to compress the vessel in order to hold it. A futher advantage of the finger structure is the ability to use laser through the fingers. The small-dimensioned gripper may be introduced through small incisions or ports to the patient. The robot is then used to perform various gripping and tool-manipulating tasks with high precision and a lower error rate than with known end-effectors. The control of the robot is achieved translating thr hand motion of an operator dirctly to motions of the gripper allowing for intuitive use.



Boaz Ben Moshe, Nir Shvalb, Dudi Asukin, Harel Levi, Amit Dvir

The current infrastructure of the Internet is not suited to simulaneous transmission of live events to millions of people. The problem is that a dedicated stream of data must be sent to every single user.
With millions of potential users, the simultaneus streams of data will easily congest the internet. The Net-HD consortium will develop a new technology that be able to increase the effective bandwidth of the network using Video adaptation, Multi Layer Cache and P2P networks without increasing the physical links.
Video on the intenet has become an integral part of the 'User internet Experience' today. The research objective is reduce BW consumption in higher layers like 'interconnect' layer and ' core' layer. Thus reduce ISP expenses.


Shamil Ploit, Boaz Ben Moshe, Nir Shvalb

We fabricated a novel approach and an implementation of virtual white cane(vwc) device.
The VWC imparts information regarding obstacles and their distance in the adjacent environment to a visually-impaired user using tactile such as stairs, sidewalks and curbs at a distance of 1-5 meters and a resolution of 1-5 inches at a rate of 100 measurements/sec.
We provide a mechnical model of the system and describe the physical models used.
Finally, we present a set of quantitative experiments performed with the VWC showing the robustness and high  performnce of the device.


Boaz Ben Mosshe, Amit Dvir, Eyal Berliner, Shamil Ploit

The company is developing proprietary TV broadcasting technologies for Wi-Fi networks that leverage the technological capabilities of smart phons & tablets WiFunTV's intial product is a software module for Pay TV platforms that enables cable/satellite companies and content providers (CP) to affordably provide video entertainment services to their mobile clients. The pay-Tv services are ramping up TV
Everywhere to compete with direct-to-consumer services such as Netflix. As a result, the momentum behind multiscreen content delivery platforms is accelerating. Some recent research forecast that revenues from multiscreen content platforms will top $21 billion in 2015. However, two showstoppers-network bottlenecks and delivery-cost make the PayTV everywhere service penetration slow. Wi-FunTV
core technology, aiming to solve problem is a point-to-multipoint sloution that allows transmission of multiple HD channels to  unlimted number of clients using standart WI-Fi Access Point.


Boaz Ben Moshe, Arkady Gorodishker, Shiry Rozenberg

Indoorgo Navigatoin System is an israeli company established in 2011, originated in Kinematics and Computational Geometry (KC&G) Multidisciplinary Labortory of Ariel Center.The company develops proprietary indoor navigation in mobile devices. The system will be used for direct communication & advertising as well as behavioral data collection on exhibition and museums visitors, supermarkets customers, etc.



Boaz Ben Moshe, Roe Yozevish, Ayal Weisman

This research address the problem of calculating the accurate of a GNSS device operating in an urban canyon, where lines of sight  (LOS) with navigation satellites are too few for accurate trilateration calculations. We introduce a post-processing position refinement algorithm, which makes use of a 3D representation of buildngs in region of interest (ROI) in conjunction with captured signals form all traceable navigation satellites. This includes weak signals originating from satellites with no line of sight (NLOS) with the device, which are usually considered not helpful in current implementations of GNSS receivers and applications. A series of field experiments confirm that approach  is applicative and that the refinement algorithm significantly improves the device's accuracy in a significant manner.


Boaz  Ben Moshe, Nir Shavl, Eyal Berliner, Aharon Gorodishker, Amit Dvir

This research is part of an Israeli consortium which covers various aspects of wireless communication for first responders. We focus on Self Organizing Mesh Networks, our research include: smart monitoring of MANET, link/rate adaptation algorithms, Forward Error Correction with limited delay. We have developed Several state-of-the-art wireless network simulations (Using NS-3) with complex GIS & propagation models.


Oded Medina, Nir Shvalb, Boaz Ben Moshe

We introduce a probabilistic algorithm (CPRM) for real time planning in the configuration space. Our algorithm differs from a Probabilistic Road Map algorithm (PRM) in that motion between pair of anchoring points takes place on the boundary of the obstacle subspace.


Oded medina, Nir Shavlb, Amir Shapiro

We a novel introduce an actuated flexible manifold.  Motion is made using artificial muscle wires.
The Forward and inverse Kinematics are calculated in the configuration space.
The manifold's manipulations are calculated in the configuration space.


Nir Shavlb, Shraga Shoval, Boaz Ben Moshe. Igor Rochaevski, Tamir Shteinberg

robot consists of  a series  of springs that are connected in a unique configuration. This configuration allows the robot to propel itself forward,  as well as to control  its propulsion direction. A soft tissue that encases all the springs creates the bell shaped body of the jellyfish robot. This tissue can capture gas within the subumbrella  volume for reducing  the mechanism weight. Three tension springs shaped in a closed ring configuration are connected parallel to each other by lateral springs.
The tension (and therefore the length) of each spring in independetly controlled using a tension control mechanism. The configuration allows for various propulsion cycles, mimicking the various propulsion modes of real  jellyfish. We base our work on a series of farms taken from a real jellyfish. Propulsion was divided into  sections  and measureaments of those sections were taken for geometry analysis. In our preliminary experiments, we achieved basic propagation in the water, and futhe motion patterns for more advanched pulsating movements are currently being develoed.Finaly we shall introduce the shape transition as an optimization problem.


oded mesina, Nir Shvalib, Boaz Ben Moshe

A C-space compression algorithm: the raw C-space is enormous and amounts to several Tera-bytses for a very poor resolution.
Obviously this cannot be analyzed with common computing force so we devised a new algorithm which simply compresses this data into a few Mega-Bytes.
This kind of data volume could be easily analyzed using common algorithms.