My name is Salem Cherenet. I am a recent graduate (May 2013) from Carnegie Mellon University with a Master of Science in Mechanical Engineering specializing in Robotics and interdisciplinary research in machine learning, data analysis and perception. I finished my undergraduate study in the Mechanical Science and Engineering department in May of 2011 at the University of Illinois at Urbana Champaign.
CAD Design, Manufacturing, Machine Learning, Artificial Intelligence, Computational Neuroscience, Robotics, Controls/Dynamics, Astrophysics: Cosmology, HCI, Mechatronics, Nano, Particle Astrophysics
Resume available upon Request.
In areas that are unsafe for humans to directly study, robotic exploration is an excellent method for collecting and interpreting data. Robots can be used for exploration; but, they require operator input about what tasks to accomplish. In order for operators to plan out optimal tasking, information about the state of
the rover is required. Panoramic imaging of the rover’s last position is useful information; however, each panorama is made up of over a hundred high resolution images, and sending this information is hard due to bandwidth limits. In this paper, we explore a way to effectively decrease the size of panoramas without losing mission important information.
Autonomous lunar landers estimate position and orientation to land and navigate safely on the moon by registering sensor data to pre-compiled terrain models. The sensor package contains an Inertial Measurement Unit (IMU), stereo cameras, a nodding Laser Imaging and Ranging unit (LIDAR) unit, and a micro-computer,
to exhibit this principle. Testing and verification of the sensor package requires collecting sensor data in a variety of platforms and terrains so that software algorithms are subjected to alternative data scenarios. This paper presents a set of methods to generate the sensor data that will be used to characterize sensor
performance for navigation and landing. Sensor data is then compared to GPS data to develop qualitative metrics for data quality and algorithm effectiveness. The implementation of these methods will yield a repository of IMU and camera data to aid in the gradual refinement of the estimation algorithms.
Team Firefleye: Hazards and accessibility problems challenge robots to be “eyes” for humans in dark planetary skylights and caves. However, robots themselves are challenged to see and document using conventional imaging techniques. Vast distances, terrainability, and total darkness limit the view point and quantity of artificial illumination that reaches the scene. The inverse square law dictates that illumination power must be at least quadrupled for every doubling of scene scale. Current approaches for dark imaging utilize cameracentered illumination which is subject to this untenable requirement. These methods result in totally dark images with little useable information.
A recent discovery is that 3D scene reconstruction is uniquely enabled by processing video obtained by hurling an illumination source through a dark scene versus beaming static light onto a scene from afar. An analogy would be filming into a dark train tunnel illuminated only by lobbing a glowing softball from the portal deep into the scene. It overcomes the classical inverse square falloff of illumination power by taking a light source to a scene rather than beaming light and has much greater mobility than a wheeled robot. The trajectory of the light’s motion is an additional basis for estimating the position of the light within the scene over time. Knowing what, where and when the light is in the scene
enables illumination assumption for computing depth, shape, surface normals and color details not otherwise achievable. This approach is known as “Firefleye”.
In the project based introduction to CAD/CAE Tools course I worked on about 11 small projects and two major group projects that provided me hands-on learning experience on how modern CAD and CAE technologies improve the three key factors of product development: product quality, development cost, and time-to-market.
The projects covered the following key topics and software packages. As part of the GEM Fellowship program I came back to SLAC for a second summer . The main goal of the project I was involved in, under the supervision of ELi Rykoff, was to find new and better way of estimating galaxy magnitude and flux errors by developing codes and algorithms to
utilized SExtractor and the newly developed LSST Data Management Software package. Please visit the Wiki for more info. For legged locomotion, dynamic balance is one of the most important things to consider. In this paper we considered a
foot placement strategy for dynamic balance by modeling
the system as bipedal linear inverted pendulum model which is based of the linear inverted pendulum model
introduced by Kajita et al. Unlike the linear inverted pendulum model, this model included double support and
single support phases. The results of the simulation showed the same force behavior as it would be expected from a linear inverted pendulum
During the summer of 2011 I participated in an internship as part of the GEM Fellowship program at SLAC National Accelerator Lab.
I worked with Dark Energy Survey group under the supervision of Aaron Roodman and Kevin Reil on several projects, which includes
Silicon Nanowires (SiNWs) are used in several applications such as converting chemical energy and
temperature gradient in to electrical energy. SiNWs could be manufactured in several different processes
depending on what final product one wants to achieve. In this experiment we are interested in fabricating
SiNWs that have separations of about 100nm and a diameter of about 40nm using the process of chemical wet etching.
We face a real issue during the last few stages of the manufacturing process when trying to transfer an alumina template
floating in buffer solution into an acetone solution then deposit it on Silicon. The purpose of this project was to
create a mechanical system that would make this process easier to carry out. To achieve this goal, a simple system was
designed and rapid prototyped for testing. The mechanical system behaved as originally predicted; however, there is still
some stability issue that needs further investigation. Working on developing and updating science codes for Community Pipeline, whose goal is to analyze
data taken by community users with DECam camera. Here is a list of topics I worked with:
As a Senior Design project, our team's goal was to design and build an abrasive/erosive wear test rig for John Deere. The rig will be used to test material performance against
wear from multiple media such as crops, soils and other small particles.
The test rig must have the capability of easily varying the media velocity/flow rate and the media/test material incident angle.
A load cell should be placed on the fixture holding the 600 g material coupon so that impact forces can be measured. The fixture
should be designed so that the material coupons can be easily changed. The machine will be restricted in size (5'x5'x door height)
and will have to be enclosed for particle containment. Parts of the machine that handle wear media should be transparent allowing for
particle interaction to be viewed. The test rig will have a hopper where the material is fed into. The hopper should be equipped with
a load cell or level indicator to indicate how much material is in the hopper. Along with delivering the media there should be an easy
way to remove it from the enclosure during or after testing. The outputs of the load cells will be read by a data acquisition system
belonging to John Deere. The students are to coordinate the sensor output requirements with John Deere. The test stand will also incorporate
a fluid delivery system that will deliver water or salt water to the rest samples to introduce corrosion. It would be desirable for the system
to utilize 120 V power, but higher voltages are available. Recently it was discovered
that the universe is expanding at accelerated rate. This acceleration could be
explained with a concept of Dark Energy. Dark Energy Survey is an imaging survey
to make precise measurement of dark energy. The first part of this paper
/presentation describes in detail the simulation done to estimate the equation
of state parameter, w and its error by comparing it with a simulation bias data.
All of the calculations were done using a Mathematica based tool created
specifically for this project. Comparing the model bias data given sigma data =
0.3 gave w = -0.716488+-0.098.
The second part of this paper focused on the method used to build a GUI that is used to analyze images from CCD cameras. The GUI is mostly stable with some bugs
still to be fixed. Carnegie Mellon University - Robotic Institute (RI)
Carnegie Mellon University - Robotic Institute (RI)
Carnegie Mellon University - Carnegie Institute of Technology (CIT)
SLAC National Accelerator Lab - Kavli Institute for Astrophysics and Cosmology (KIPAC)
Carnegie Mellon University - Robotic Institute (RI)
SLAC National Accelerator Lab - Kavli Institute for Astrophysics and Cosmology (KIPAC)
Nano-Photonics and Nano-manufacturing -Illinois Scholars Undergraduate Research Program (ISUR)
Dark Energy Survey Data Management Group (DESDM)-The National Center for Supercomputing Applications (NCSA)
Design and Build of an Abrasive/Erosive Wear Test Rig-University Of Illinois at Urbana-Champaign (UIUC) and John Deere
Dark Energy Survey via the technique of Counts of Galaxy Cluster. -Fermi National Lab
Salem H. Cherenet
Email: salemch@cmu.edu
Email: salemch@salemch.com
Social Links
Linkedin ...salemc
Twitter ...salemch
Youtube ...selamawiw
Google Plus ...salemch