Puckett lab summer 2016

The two systems above may seem very different  (one is a granular system made up of thousands of photo-elastic disks and the other is a school of several fish), but they are similar in very interesting ways. Our lab studies soft condensed matter systems which are composed of many small parts that exhibit observable large scale properties. How do local interactions influence global properties?  How do these systems transition from disordered to ordered states? How do they respond to external conditions such as strain or stress?


Collective Animal Behavior: Visual Perturbations of Laboratory Schools: (Julia)

Collective animal behavior is observed across many scales and organisms. While different species have different environmental and social pressure, many similar structures are found across the animal kingdom.  Clearly, these group behaviors are a strategy nature likes to use which helps individuals work together and respond to their environment in a rather general way. These structures emerge from from interactions between individuals. A common model for collective animal behavior involves zones of repulsion, orientation, and attraction. Individual animals seek to maintain a safe distance from and align with their neighbors. While several models exists and qualitatively mimic nature, they do not capture the dynamics well. We address this issue by obtaining and analyzing precise spatiotemporal data of the response of laboratory schools of rummy nose tetra to environmental stimuli. Using high-speed video, we track individuals and investigate how emergent properties arise from individual behaviors.

The trajectories of each fish in the school provides us with a wealth of data to apply statistical mechanics.  Some of the metrics we are looking at are: density, speed, volume, social force magnitudes, and rotational and translational order parameters. We also use our data to study phase transitions of schools and examine how they go from a disorganized state to a mill (shown below).

So that we do not disturb our fish, our camera captures an infrared image of the fish — which the fish cannot see.  We calibrate the camera to correct for lens distortion and transform pixel coordinates into world coordinates.  The water in our experimental tank is temperature controlled, pH controlled, current controlled, and shallow to create a pseudo-two-dimensional system.  This provides a powerful tool to determine how individual interactions connect with the dynamics of the school. 

Using a projector, we perturb groups of rummy nose tetra with dynamic light gradients (as seen in the video below).  Rummy nose tetra are a social tropical species that naturally avoid areas of high light intensity and move towards darker regions. We can therefore manipulate our laboratory schools and observe their collective response.  The noise image consists of two images superimposed: a moving gaussian blob (seen easily in the figure to the left), and a multi-frequency noise field.  We use this to examine the performance of the school in tracking the darkness.

Below is a video of the noisy image that is projected onto the experimental tank with an image of the fish (tracked by the infrared camera) overlay in red/orange.

Day-to-day, our work largely consists of caring for populations of rummy nose tetra, maintaining and improving our apparatus, and writing and developing the computer programs that we use to stimulate experimental schools, process data, and analyze data. There are several challenges that arise from working with such a complex living system. Collecting and analyzing data can be very time consuming because we collect very large amounts of data. Throughout the rest of the summer we will continue exploring the interesting physical properties of social animal groups.

Simulations of Collective Animal Behavior:  (Aawaz)

Our fish are negatively phototactic — which means they like to be where its dark.  But how do fish find the darkest spot?  A previous study showed that fish cannot see the gradient of their environment (getting lighter or darker) only the absolute light level.  Yet, somehow, collectively the group can find the darkest spot.  We continue this work and further examine the mechanisms of the school that lead to this emergent sensing.

We examine this effect using a simulation of our school of fish in a noisy environment. The rules of the simulation are simple: don’t get too close to others, move toward others if too far away, align with others that are somewhere between, and slow down the darker the background.  Implementing this simulation using a GPU is not so simple.  

What’s amazing is that the fish have no way of detecting the gradient (slope) of the darkness — so how do they find the darkest spot?  This is an example of emergent sensing: see the video below of one of our simulations.

Granular Materials: (Alex)

Granular materials are the second most utilized substance in industry next to water. Despite this, granular materials are still not well understood and models are still lacking predictive power that would be lifesaving (earthquakes, avalanches).  One promising approach is using statistical mechanics to examine these systems as an ensemble of particles.  In the PuckettLab, we analyze a two dimensional granular system composed of plastic cylinders (PSM-4).   But this is special plastic, when view between crossed polarizers, a fringed image is formed that is unique to the force on the particle.  Below left is an image of our granular material and to the right is an image of the stress network.


Much of the work this summer has been in the vein of error elimination and apparatus automation. To mitigate parallax error a dual camera system was constructed so as to stitch together the low parallax zones of two images. The Parallax Inhibiting Camera Setup (PICS) has hence been deconstructed replaced with a single camera and mirror setup which increases the visual path length enough that the parallax angle is acceptably small. This Focal Length Extension Mechanism (FLEM) has the added benefit of decreasing computer processing time as standalone images suffice for our data collection.

To switch between particle finding and stress network images the analyzer was previously rotated in front of the camera by a stepper motor. However due to the nature of the motor the analyzer partially undershoot or its ideally polarizing location. This had the effect of randomly decreasing the fidelity of stress images with non-parallel transmission axes. A new Automated Analyzer Actuation Apparatus (AAAA) has since been constructed  which linearly moves the analyzer.  This contraption does have uncertainty in its polarized and unpolarized endpoints however this is made negligible with a large enough analyzer as the transmission axes will always be at the same angle between data points.   

The compress-image-polarize-image-unpolarize-shear-image-polarize-image-unpolarize-shear-etc process has been successfully automated with a LabView Virtual Interface.  The particle finder and stress network analysis programs are being refined.  









Charged on 3 Counts of AsSalt

In the Andresen lab we research biophysics, specifically involving DNA. Fun Fact: DNA needs a trivalent cation to condense (hence the title). Here is some in depth information about our projects:

My name is Savannah Miller and I am a rising senior with a BMB major and a Physics minor. I worked in Professor Andresen’s lab last year on a similar project and enjoyed it so much that I decided to work here again this summer. I decided to work in Professor Andresen’s lab because I was attracted to the interdisciplinary nature of the biophysics research. I felt that this lab would give me the chance to apply my background in both biochemistry and physics and learn how to approach research questions from a wide variety of angles. My project this summer is an exploratory investigation into how cationic gold nanoparticles interact with DNA. I am collaborating with Professor Thompson’s lab who are providing me with gold nanoparticles and their expertise. Gold nanoparticles are nanoscale gold particles that are stabilized by charged organic molecules like CTAB and citrate. We have been using CTAB stabilized gold nanoparticles that are positively charged and thus will electrostatically attract the negatively charged phosphate backbone of the DNA.


Flow Reactor used to make large quantities of CTAB AuNPs

The DNA we have been using is calf thymus DNA which ranges from 8-15 kb in length. We expect the nanoparticles to act as a condensing agent for the DNA and the DNA to maybe even form histone-like structures around the nanoparticles. Unfortunately the addition of the long DNA caused instability and aggregation of the nanoparticles so we are now planning on using smaller DNA that will be less likely to incite aggregation. To obtain smaller DNA we have decided to shear down the DNA we have using a sonication protocol mentioned in a paper. The protocol uses a probe sonicator (see below) to instigate cavitation, the formation and collapse of microbubbles that together spread damaging shockwaves through a solution, in order to shear calf thymus DNA down to at least 800 bp.

Using an old probe sonicator from the biochemistry lab, we were able to run the protocol listed in the paper. Gel electrophoresis shows that the sheared DNA solution contains a variety of fragment ranging from 1kb to far below 0.5 kb. This is promising because it means we have successfully sheared the DNA and obtained fragments that may be able to interact with the nanoparticles without stimulating aggregation.

We have run out of CTAB gold nanoparticles, which are more difficult to make than citrate gold nanoparticles, so I am going to start using the citrate ones. Citrate is naturally negatively charged which would repel the negatively charged phosphate backbone. Instead, I plan to layer lysine onto the nanoparticles on top of citrate. Lysine is positively charged at physiological pH so it will create the correctly charged surface for DNA electrostatic attraction.

My name is Sarah Hansen and I am a rising senior, majoring in physics and minoring in chemistry. For the past two summers, I have worked with Professor Andresen, whose research involves biophysics. Last summer, my project focused on ion-counting studies around arrays of DNA strands, but this summer my project involves nucleosomes.

Because DNA strands are negatively charged (due to the phosphate groups), they need to wrap around histone proteins in cells to condense. Several histone proteins will form a “bead on a string” type conformation, eventually forming chromosomes. Though, the exact structure and electrostatics of these interactions is unknown. The goal of my current project is to investigate the electrostatics of DNA-wrapped histone proteins, perhaps using ion-counting techniques like last summer.

Before I study the properties of DNA-histone complexes though, I must first make a sample of them. I began my summer with 50 mL of whole chicken blood (see image below), which contains plasma, white blood cells, and red blood cells. Unlike humans, the red blood cells in birds actually carry DNA, so the goal was to isolate the red blood cells, and then further isolate the nuclei.

Components of Whole Blood-1


Through a series of biological and chemical techniques, I have been working on isolating the nucleosomes from the whole chicken blood (see image below for an example of spinning down my samples to achieve nuclei). Although I have not reached my end goal yet, I am close to making my sample. Throughout the rest of the summer, I will work on isolating the nucleosomes, as well as running electrostatic experiments on them.

Sample image

In summary this is what research is like in the Andresen lab:



CRISPR Crew Summer 2016

In Dr. Shariat’s lab (aka the “CRISPR Crew”), we’re all rising sophomores and we have three main research focuses: the function of the CRISPR-Cas system, the applications of the CRISPRs for sub-typing, and the use of the phage therapy in controlling plant pathogens. We study two closely related bacterial pathogens: Salmonella enterica and Erwinia amylovora.


Bhaya D. et al 2011

Clustered Regularly Interspaced Short Palindromic Repeats with their associated cas genes (CRISPR-Cas) are found in 48% of sequenced bacterial genomes. CRISPR-Cas systems are typically used as a bacterial defense mechanisms against foreign DNA such as bacteriophage and plasmids.

CRISPR-Cas systems contain cas genes, typically located next to a CRISPR-Cas array. These arrays consist of spacers and direct repeats, which code for RNA that guide a cas nuclease to foreign DNA injected by phage, thus targeting the foreign DNA for destruction. The spacer sequences are generally derived from foreign DNA elements. CRISPR-Cas systems are adaptive: as the bacteria comes into contact with the new phage, spacers can be added to the array that are specific to that particular phage.

Function, Function, Function!

Jake and Kaelea

Pellet, what pellet?


Jake and Kaelea: Most studied Salmonella genomes have been shown to contain a CRISPR-Cas system, with eight cas genes and two arrays, CRISPR 1 and CRISPR 2. CRISPR 1 is adjacent to the cas operon. Evidence suggests that the Salmonella CRISPR-Cas system is not used for immunity, as there is a lack of spacer acquisition and that only a few spacers actually match that of phage sequences. Interestingly, Salmonella has maintained the genetic integrity of its CRISPR-Cas system as seen by the lack of mutations within the cas genes and spacers.  We spend much of our time in lab growing different types of Salmonella and performing assays on components of the CRISPR-Cas system. Jake can now do bacterial RNA preps in his sleep!

Celine: When we can drag her away from Phage work, Celine is looking for novel CRISPR-Cas systems in other bacteria.

Using CRISPRs as Molecular Fingerprints

Dorothy and Stefani:

Dorothy and Stefani are using a subtyping technique called CRISPR-MVLST to identify specific strains of Salmonella found in birds.  CRISPR-MVLST uses sequence information from both CRISPR1 and 2 plus two virulence genes.  Spacer differences in the CRISPR arrays and single nucleotide polymorphisms (SNPs) in the virulence genes are used to define and separate individual Salmonella strains.


Photo Jun 21




For analyzing the CRISPR arrays, we use a macro, which only displays the spacer sequences (the bits of DNA that come from phage and which are unique).  The macro generates an image like the one to the left where different colored boxes represent different sequences.


Dorothy is investigating a particular type of Salmonella that is often found in chickens and occasionally in humans.  She’s been looking at human isolates of Salmonella and is finding some cool CRISPR patterns!

Stefani is a rising senior at Biglerville High School and is working in our lab as part of new program “Bridging the Furrow Ag-Research Training” for URM high school students.  She’s been looking at strain differences in Salmonella isolated from different songbirds from different states, including the American Goldfinch and the Brown Headed Cowbird.



Cameron’s project focuses on the use of the Salmonella CRISPR array as a means of identifying what strains are present in a complex sample of Salmonella strains. Each strain has a slightly different composition of known unique spacers. By identifying these spacers, they are able to identify different Salmonella strains.  Last week we traveled to to collect samples at a poultry facility: many chickens and lots of chicken poop. Check out our cool suits though!

Photo Jun 13

All suited up in PPE and ready to go!       

Photo Jun 13 copy

Dorothy and Cameron collecting chicken poop


Photo Jun 16

Salmonella grows as distincitve black colonie

And Cameron found lots of Salmonella – he’s doing PCR analysis of the CRISPRs as we write!


Phage hunting

Celine and Caleb:

Bacteriophage can be found anywhere that bacteria are found and have been studied for over 100 years.  They are a specific type of virus that infect and kill bacteria and as such, can provide an alternative therapy to antibiotics in treating bacterial infections.  Celine and Caleb have been isolating and purifying phage against Erwinina amylovora, a bacterial pathogen that infects apple trees.

Collecting Fire Blight Samples


Here’s some local trivia for you: did you know that Pennsylvania is the fourth largest producer of apples in the United States? And that 75% of Pennsylvanian apples are grown right here in Adams County? Many apple orchards are within a 15-20 minute drive of campus – take a drive and see them, they’re beautiful!


In the lab, Celine and Caleb perform different assays to purify phage, including a plaque assay as shown below.  They’re getting really good at this – check out all those plates!



A phage plaque assay                                              


Photo Jun 08

Many, many plaque assays


Photo Jun 10

CRISPR-Crew, Summer 2016



Investigating Epigenetic Gene Regulation in Aspergillus nidulans

Background on…. the Cell Cycle, Epigenetics, Aspergillus, snxA, etc. 

In Dr. James’ lab we conduct cell cycle research in Aspergillus nidulans, a common fungus that also happens to be our model organism of choice.  More specifically, we study the regulation of these cell cycle genes, something that we promise to discuss later in the blog.

The cell cycle simply (or rather complexly actually) refers to the normal cycle of growth and reproduction whereby a cell divides to form two identical daughter cells, a process called mitosis.  During this cycle many well orchestrated and tightly controlled process occur, such as the dissolution of the protective membrane surrounding the nucleus, chromosome condensation (10-100 fold), and microtubule breakdown and reassembly.  Following microtubule breakdown and reassembly into a spindle apparatus, the complex moves into the nucleus, attaches itself to the chromosomes, and pulls them apart to form two nuclei that contain identical genetic information, forming the basis for the two new daughter cells.

As you can imagine with a process this complex and essential to eukaryotic life, there are many key players, the majority of which are different genetically encoded proteins.  In A. nidulans, regulation of the cell cycle depends upon a group of proteins that are highly conserved across all eukaryotes, making the fungus an excellent model organism for cell cycle research.  What our group is especially focused on is understanding the regulation of a gene called snxA, which was originally identified as a suppressor of a second gene involved in mitotic induction (here is the cell cycle tie in).  The way that this works is that when the second gene (named nimX) is mutated, mitotic entry is prevented and the cell dies.  However, mutations in the snxA gene, named snxA1 and snxA2, are able to alleviate the mitotic defect conferred by nimX mutations, forming the basis of an interesting relationship (another way to think about this is that when snxA is mutated, the nimX gene is no longer needed for mitosis).  After discovering this interesting genetic interaction, the James lab has embarked on a journey to understand the function and regulation of the snxA gene in Aspergillus.  

What we have worked hard to reveal thus far is that the snxA mutations, snxA1 and snxA2, are not caused by changes to the snxA DNA sequence, and are actually the result of stable epigenetic modifications to the snxA DNA-protein complex.  Simply put, the term epigenetics encompasses the changes above and beyond the DNA level, which determine whether a given gene is activated or repressed.  In this way different chemical modifications added to the DNA and DNA associated proteins (termed “chromatin,” the DNA + protein architecture).  In A. nidulans, however, the only source of epigenetic variance stems from the acetylation and methylation of the various histone proteins that comprise the majority of proteins found in chromatin.  Through the strategic positioning of acetyl and methyl groups on these histones, genes can be more actively transcribed, as the addition of these groups negates the positive charge of amino acids in the histones, pushing the negatively charged DNA away from the chromatin.  This process ‘opens’ up the DNA, allowing more active transcription of the genes by the cellular machinery.

Screen Shot 2016-06-20 at 9.56.19 AM

Quick diagram showing the acetylation of histone proteins (on the left) marks active gene expression by allowing transcription factor (TF) access to DNA.  Complex patterns of histone modifications occur in eukaryotes and can also act to compact the DNA and repress gene expression (right).

More Specifics on the Projects…

My project (Hi I’m Sarah) is very closely related to the epigenetic investigation of snxA, and involves investigating a suppressor of the snxA2 defect called set2. This gene in A. nidulans encodes protein SET2 that has been characterized as a histone H3K36 methyltransferase and normal function involves repressing transcription with the introduction of 3 methyl groups at lysine 36 of histone H3. Previous work using Western blotting, where proteins are identified using specific antibodies, has shown that a set2 gene with a point mutation (set2-sup59) in a snxA2 background has a high level of tri-methylated H3K36 protein. This is unusual if we assume that the set2 gene with a point mutation has eliminated the proper function of the protein. My project this summer will involve investigating this finding again but rather than using a point mutation, we will use the same Western blot technique to see if a ∆set2 (a complete deletion of the set2 gene) mutant will have the same unexpected high level of tri-methylated H3K36. This process hasn’t started yet, so in the meantime I will continue to assist other projects that Dr. James has in the works in the lab.

I (Morgan) am also largely focused on snxA this summer and my goal is to find what we consider the “regulator of snxA”—a gene that is genetically but not physically linked to snxA that controls snxA function from a location elsewhere in the genome. Previously, we detected this exact linkage (two genes are said to be “linked” if they are in close proximity to one another on the same chromosome) in an area on chromosome I. After whole genome sequencing of several snxA1 and snxA2 mutants, the only gene on chromosome I with a mutation (we believe the regulator of snxA gene must have DNA mutations in snxA1 and snxA2) in every mutant was AN6263. These gene codes for an AAA-ATPase, a protein involved in powering a wide range of processes throughout the cell by translating the energy released through ATP hydrolysis into conformational changes on various substrates. The AAA+ proteins make up one of the largest families of proteins found in living cells, with one of their functions including the regulation of gene expression—which is exactly what we expect is happening with snxA mutants. AN6263 in particular doesn’t share many domains with other ATPases known in A. nidulans and its close relatives, which means there’s a possibility we are working with a completely novel gene, which is exciting in-and-of itself.

My goal in these first few weeks has been to make copies of both the wild-type and mutant versions of the AN6263 gene in order to answer three questions: 1, will cloning the wild-type gene into a snxA mutant restore the wild-type phenotype; 2, will deleting the gene from a wild-type strain create a snxA mutant; and 3, will inserting the gene containing the point mutation that we discovered during whole-genome sequencing turn a wild-type strain into a snxA mutant? We’ve already used PCR to create the necessary genes, isolated and purified them, and transformed them into A. nidulans. Now all that is left is to wait and see which transformants display the snxA mutant phenotypes, and from there we can determine whether or not AN6263 is indeed the gene we’re looking for!

Now you should be mostly caught up on everything going on in the lab! Thanks for reading!

A New Brood

Hello again! I am Will Ueckermann, and this is the account of how I have spent my second summer in the Hiraizumi lab so far. This summer’s research has been a continuation of last summer’s, where I am working on a few projects. Our research focus is on the functional and sequence variation of the 5′ untranslated regions (5’UTRs) of genes in Drosophila melanogaster, the common fruit fly. Our particular focus is on the Dipeptidase B gene (Dip-B), which codes for one of the three dipeptidases found in D. melanogaster. Dipeptidases hydrolyze dipeptides, compounds that consist of two amino acids bound together. They are found in all organisms, from bacteria to humans, making them a good model for examining mechanisms of gene regulation. While 5’ UTRs do not code for any part of the protein, they are still very important due to their impact in protein translation. The DNA sequence within 5’UTRs can contain regions referred to as pseudointrons that are often spliced out. Although the study of pseudointrons is still in its infancy, it is known that 35% of all human transcripts have them.

We use two strains of D. melanogaster in our lab: CL55 and NC25III. CL55 exhibits a wild-type level of DIP-B enzymatic activity while NC25III expresses significantly low enzymatic activity. The ongoing project for the past few summers has been to determine the molecular, biochemical, and genetic basis for this difference.


One of the bottles used to maintain a population of a fly strain.

My experiments utilize three general procedures: nucleotide extraction and purification, polymerase chain reaction (PCR), and gel electrophoresis. Nucleotide extraction is the isolation of DNA or RNA from samples of specific fly strains. This entails collecting a certain number of flies, grinding them up into a slurry, rupturing their cells and then cleaning out all of the other cell fragments, leaving only DNA or RNA in the prepared sample. The nucleotide samples can serve as material for several experiments involving PCR.


Anesthetized flies, ready for sorting for nucleotide extraction.

PCR is used to amplify a specific subset of the nucleotide sample which represents a nucleotide sequence that can be specifically studied, such as the coding sequence of a particular gene. The procedure involves a pair of primers to target the desired sequence to be amplified, and a DNA polymerase enzyme to synthesize a new sequence defined by the boundaries of those primers. In effect, by repetitive cycling through the targeted synthesis reactions, millions of copies of the desired DNA sequence can be generated.


A thermocycler used for PCR, with the lid open, ready for an experiment.

Gel electrophoresis of the PCR products is conducted to confirm fidelity of the reaction and presence of the predicted products. This involves subjecting the nucleic acid sample to an electrical gradient through a flat agarose matrix. Because DNA is negatively charged under electrophoresis conditions, molecules migrate through the pores of the agarose gel as a function of their size. DNA molecules are stained with a fluorescent dye and bands of differing sizes can be visualized against DNA reference markers of known size.

Gel Chamber

Electrophoretic separation of PCR products in progress. The orange dye front migrates ahead of all nucleotide samples and serves as a visual indicator for completion of electrophoresis.

Perhaps our most exciting progress is the discovery of dipeptidase B mRNA isoform E. When the Dip B gene is transcribed into messenger RNA, four mRNA isoforms that differ in sequence and length are produced: isoforms A, B, C, D. Isoforms B and D result from different sites for initiation of transcription, while isoforms A and C share the same transcription initiation site. The only difference between isoforms A and C is the size of their pseudointrons that are spliced out to produce these mRNAs. These two isoforms have the same 5’ splice junction for the pseudointronic splicing, but they differ in the location of the 3’ splice junction, leading to differences in molecular lengths that can be detected and identified by agarose gel electrophoresis.


Differences in size and initiation site for the five Dip-B mRNA isoforms.

The pursuit of isoform E began last summer was based on an observation of an unexpected PCR product generated by a previous student who had designed a PCR primer combination to amplify portions of isoforms A and C. In addition to the two predicted PCR products, two other molecules were generated, one of which was unspliced RNA, perhaps a pre-mRNA for isoforms A and C. However, there was another PCR product that was larger in size than the other two isoforms, but smaller than the unspliced RNA. After hypothesizing that this may have been an uncharacterized and unreported isoform, I began designing primers to specifically amplify it from samples of Drosophila mRNA. Working from the assumption that the possible isoform shared its 5’ splice junction with isoforms A and C, I designed a series of primers that bridged the gap defined by the pseudointron sequence, which would specifically amplify only this putative isoform. After several candidate primers were tested, one primer was found that generated a PCR product from total mRNA in the same size range as it was designed to. This was our first real support for the presence of isoform E.

Fast forward to this summer, and we are now working to isolate isoform E by PCR amplification so that the resulting product could be sequenced. The first set of PCR experiments was completed last week to confirm the repeatability of last year’s results and to eliminate the possibility of genetic contamination. The experiments were also conducted to see if the putative isoform was present in both males and females of both strains. While each PCR experiment did yield a product, there were some interesting results, such as PCR products from female samples being smaller than those from male samples. The difference was repeatable under varying electrophoretic conditions. In order to determine the nature of this difference, the PCR products will need to be sequenced.

Another project that we have been working on is the isolation of the gene sequence of Dip-B from genomic DNA of CL55 and NC25III. Numerous attempts to PCR amplify the genomic gene sequence were made over the course of a year, involving many primer combinations with different DNA polymerases. This summer, I finally seem to have found a primer combination that can amplify the entire gene sequence from genomic DNA using Taq DNA polymerase. Interestingly, the Dip-B gene sequence for NC25III seems to be smaller than that of CL55, suggesting a deletion within the Dip-B gene for NC25III. To determine if the deletion is associated with the exonic or coding sequence, these PCR products will be isolated and sequenced.


Gel image of PCR amplified Dip-B gene sequence from genomic DNA.  Lanes one and three correspond to CL55 PCR product, and lanes two and four correspond to that of NC25III.

How does a robot localize itself?

One of the fundamental problems of robotics, localization, can be understood as the robot asking the question ‘Where am I now?’ Given a map of the robot’s surrounding environment and models of the robot’s motion and sensors, determine where the robot is.  Imagine that there is complete darkness in a familiar room and you only have sense of touch.  Touching a familiar object help you localize in the room.  As you leave that familiar object and walk across the floor, your uncertainty grows until your senses and recollection of the room help you localize.   The Kidnapped Robot Problem is similar.  Imagine that a robot with a map and behavioral models is placed in an unknown location with an unknown heading.  How might you program it to localize, i.e. determine it’s current state or ‘pose’.

Monte Carlo Localization (MCL) is an algorithm that begins with a set of random hypotheses about where the robot might be all over the map and in any heading.  As the robot moves and senses, a Darwinian survival-of-the-fittest process tends to multiply the most likely hypotheses and tends to kill off the least likely, gradually evolving a cloud of hypotheses where the center, i.e. average,  is the most likely robot position and heading.  Each of these hypotheses is called a “particle”, as this technique derives from a technique called Particle Filtering.

The algorithm runs like this:

1. Initialize set of randomized particles, in other words hypothesis about the robot’s location.

2. Gather data about the physical environment including sensor information and motion information.

3. Look at each particle and assign a weight based on how well that it fits with the data gathered.

4. Resample a new set of particles according to current particles. Particles with higher weights populate more particles in the next round.

5. Then repeat from step 2.

There are two ways to calculate the final hypothesis of robot pose. One is to calculate the weighted sum of all the particles. The other is taking the particle with the greatest weight as the result.

One hazard of failure in MCL is when the population of hypotheses lacks diversity and misses the mark.  One can always restart the process, but there are improvements to MCL that can strike a balance between explorative diversity of hypotheses and exploitative focus on best hypotheses.  By monitoring the quality of the population of hypotheses, i.e. the particle “weights”, over time, one can introduce greater/fewer new hypotheses to the population.


Zuozhi Yang

How long can a neutron live?

Hello, my name is Fangchen Li. This summer I’m working with Professor Crawford on a Neutron lifetime research. We are part of a research group. This group is trying to design an in-beam experiment to measure neutron lifetime.
Neutrons in stable nuclei can exist forever; a free neutron can last for about 15 minutes before decaying to a proton, an electron, and an antineutrino. People’ve done two experiments to measure this quantity. One is the bottle method. A certain number of neutrons are placed in a container. After a fixed time, people open the container and count how many neutrons are left. And another method is called the beam method. As a beam of neutrons passes through a particular volume of space, the number of neutrons decay to a proton can be measured. Then we can calculate neutron lifetime from this number.

But the beam and the bottle experiments do not agree. Our group is trying to improve the beam experiment to get a better measurement. Professor Crawford and I are working on developing Matlab programs to simulate the beam experiment. We use Monte Carlo Method to simulate neutrons. Then we trace each neutron as it goes through the experimental apparatus. In a computer simulation, we can easily find out how the results change as we vary the experiment setup.

So what am I doing every day? I sit in front of my computer, write code, and test my code.