MEAM Seminar Series Summer 2016
For Spring 2016 Seminars, click here.
Seminars are held on Tuesday mornings beginning at 10:45 am in Towme 337 (unless otherwise noted).
To be added to the MEAM Events mailing list (which sends notifications regarding all departmental seminars and events) please email us at email@example.com.
Monday, June 13
PhD Thesis Defense
James Keller, PhD Candidate, University of Pennsylvania
Advisor: Vijay Kumar
11:30 am, Levine 307
"Path Planning for Persistent Surveilance Applications Using Fixed-Wing Uninhabited Aerial Systems: A Constructive Solution Using B-spline Parametric Polynomials"
Persistent surveillance is an important application that can be undertaken by autonomous aerial vehicles. However, persistent surveillance path planning algorithms are particularly degraded when vehicle maneuverability and agility constraints render simple solutions unfeasible, especially when the surveillance objective drives the importance of optimized paths. Researchers have developed a diverse range of solutions but few directly address dynamic maneuver constraints.
The current state of this technology is summarized and requirements for path planners in this context are compiled. A wide variety of approaches is discussed to enable a down-selection to a suitable technology for development of a practical planner.
We present a two stage solution that combines the graph search techniques with parametric polynomial curves.
We show that B-spline parametric polynomial curves provide a foundation on which detailed plans can accommodate path constraints while following the framework developed through the graph search. In particular, quartic splines are shown to be the minimum polynomial basis order with which to produce paths consistent with air vehicle equations of motion. The B-spline format encodes paths in a minimal representation, which permits direct integration with standard autopilots. We demonstrate the approach can accommodate obstacles and collision avoidance check to provide a practical solution.
Wednesday, June 15
Denise Wong, PhD Candidate, University of Pennsylvania
Advisor: Vijay Kumar
2:00 pm, Towne 337
"Actuation, Sensing, and Control for Micro Bio Robots"
Microscale robots have applications in micro-assembly, directed drug delivery, microsurgery and high-resolution measurement. To accomplish these tasks, the robot must be able to precisely position in the environment, take local measurements and use these measurements to make decisions. Despite significant advances in microscale imaging, measurement and micro- and nano-fabrication, there are no off the shelf components for building microscale robots. Biological organisms dominate the microscale environment; by looking to biology, we find analogues motors, sensors and processors. To develop micro robots, we take inspiration from biology and combine organic and inorganic components to create programmable micro machines. Specifically, we integrate synthetic micro structures with single-cell biological organisms to provide un-tethered on-board actuation to move robots. We also integrate synthetically engineered cells as sensors to create a system that can detect, store and report information about local environmental changes. At a slightly larger scale, macroscopic structures embedded with magnets are used to direct the assembly of passive inorganic engineered structures, which have the potential to be synthesized with biological components. These magnetic robots can be used to program the assembly of passive building blocks to create complex structures. Our results make contributions toward actuation, sensing and control of autonomous micro systems and leads to the development of swarms of micro robots with a suite of manipulation and sensing capabilities.
Tuesday, July 19
Yijie Jiang, PhD Candidate, University of Pennsylvania
Advisor: Kevin Turner
"Characterization of Nanoscale Adhesion and Wear at UNCD–PMMA Interfaces"
10:45 am, Raisler Lounge
Atomic force microscopy (AFM) is a powerful tool for high resolution measurements of surfaces as well as for tip-based nanomanufacturing. In these processes, a sharp tip mechanically interacts with the surface of a polymer film coated on a substrate. Understanding the nanoscale tribological behavior of the tip-polymer contact, including adhesion and wear, is important for the control of these processes. In this work, AFM-based adhesion and wear experiments between ultrananocrystalline diamond (UNCD) AFM tips and polymethyl methacrylate (PMMA) films were performed. The experiments were coupled with finite element analysis (FEA) to understand the mechanisms of nanoscale adhesion and wear.
In the adhesion studies, the properties of the adhesive traction-separation relation between a UNCD AFM tip and PMMA sample were characterized using a novel AFM-based method that combines pull-off force measurements with characterization of the 3D geometry of the AFM tip. Using the pull-off force data, the measured 3D tip geometries, and an assumed form of the traction-separation relation, specifically the Dugdale and 3-9 Lennard-Jones relations, the range, strength, and work of adhesion of the UNCD-PMMA contact were determined. The assumptions in the analyses were validated via FEA. Both forms of the traction-separation laws result in a work of adhesion of approximately 50 mJ/m2 and the peak adhesive stress in the Lennard Jones relation is found to be about 50% higher than that obtained for the Dugdale law. The main contributions of this study are detailed measurements of the adhesion at UNCD-PMMA interfaces and a novel technique for determining the range of the adhesion.
Measurement of wear of polymer surfaces at nanoscale is difficult as inaccurate measurements can result from the debris produced by the tip-sample interactions. Furthermore, Archard's wear law generally does not describe wear behavior at small scale, making interpretation of data an open question. Nanoscale load-controlled AFM wear experiments were performed on electron beam patterned PMMA structures, which had gaps that allowed debris to be captured. The contact stress distribution at the tip-sample contact was calculated by FEA based on the measured 3D geometries of the tips and the applied loads. Both line wear tests and raster wear tests were performed. The results of the line wear tests indicated that the material removal rate as a function of stress is well described by a recently proposed transition state wear mechanism. The activation energy and effective activation volume were obtained from line wear and then used to predict the volume loss in raster wear tests. Despite of many differences between experimental configurations in the line and raster wear experiments, the predictions using parameters obtained from the line wear tests were in good agreement with the raster wear experimental results.
Thursday, July 28
Justin Thomas, PhD Candidate, University of Pennsylvania
Advisor: Vijay Kumar
"Autonomous Grasping and Perching for Micro Aerial Vehicles"
10:45 am, Towne 337
The ability to maneuver Micro Aerial Vehicles (MAVs) relative to specific targets and to interact with the environment could benefit society by assisting with dangerous jobs, providing useful information, and improving the efficiency of many tasks. For example, accurate relative positioning would allow for close inspections of bridges, cell towers, rooftops, and water towers. Aerial manipulation could improve or enable precision farming, construction, repairing structures, transportation of objects, automated recharging, environmental sampling, and perching to turn off motors and reduce power consumption.
The prevalence of commercially-available MAVs has risen rapidly, but platforms are currently limited to sensing and data collection tasks, and none are capable of physical interaction with objects. Thus, there is a need for solutions empowering aerial robots to closely track, grasp, perch on, and manipulate specific objects of interest.
This work is focused on facilitating interactions between quadrotors and surrounding objects. We first explore high-speed grasping, enabling a quadrotor to quickly acquire an object while moving at a high relative velocity. Next, we discuss planning and control strategies, empowering a quadrotor to perch on vertical surfaces using a downwards-facing gripper. Then, we present an overview of current vision-based approaches and discuss challenges for vision-based perching and aerial manipulation. Finally, we show how a quadrotor can use a single camera and IMU to perch on a cylinder without an external motion capture system.
Tuesday, August 2
Liang Xiaojun, PhD Candidate, University of Pennsylvania
Advisor: Prashant K. Purohit
10:45 am, Towne 337
"A Fluctuating Elastic Plate Model for Lipid Membranes and Graphene"
The function of a membrane is strongly influenced by its specific two-dimensional structure. It is crucial to study how its mechanical behaviors could be mediated by thermal fluctuation, especially in the out-of-plane dimension. Mode analysis, molecular simulation, and Monte Carlo simulation of triangulated membranes is usually applied to this sort of problem. We propose a new semi-analytic approach that is less computationally intensive. We first discretize the membrane using triangular finite elements and express its energy as a function of nodal displacements, to quadratic order. Then we compute its partition function and other thermodynamic properties using Gaussian integrals. We examine the dependence of membrane projected area on applied tension, and the dependence of membrane free energy on geometry, spontaneous curvature, boundary conditions, and tension. As new applications we compute the fluctuations of the membrane of a malaria infected cell and analyze the effects of boundary conditions on fluctuations. We also compare our calculation’s efficiency and accuracy with some simulations. We extend our technique to solid membranes, such as, graphene. We carefully examine the effect of coupling between in-plane and out-of-plane displacement on thermal fluctuations. Again, we recover well-known results for the scaling of the fluctuations with membrane size, but we show that the fluctuation profile strongly depends on boundary conditions and type of loading applied on the membrane. We quantitatively predict the dependence of the thermal expansion coefficient of graphene on temperature and show that it agrees with several experiments.
Thursday, August 4
Tolga Ozaslan, PhD Candidate, University of Pennsylvania
Advisor: Vijay Kumar
2:30 pm, Towne 337
"Micro Aerial Vehicles for Safe Autonomous Inspection of Critical Infrastructure"
In the last decade, multi-rotor Micro Aerial Vehicles (MAVs) have attracted great attention of robotics researchers. Due to their agility, maneuverability and simple design with affordable costs, MAVs are platforms cut out for real-life applications. For these reasons MAVs are becoming more common in civilian applications such as maintenance of power-lines, cell-towers and precision agriculture.
In particular, we are interested in MAV platform and sensor fusion algorithm design to facilitate inspection and maintenance of large infrastructures such as dams, locks and penstocks. While GPS offers an easy solution for outdoor autonomy, using on-board sensors is the only solution for autonomy in constrained indoor environments such as penstocks. Furthermore, most off-the-shelf MAVs equipped with basic sensor packages do not qualify for operating in such challenging environments. This work presents hardware and software solutions to the challenges specific to on-board pose estimation and navigation of MAVs operating in pitch dark, symmetric confined spaces that lack geometric and visual cues. The proposed system can be used to periodically inspect and map the structure to detect features that might indicate potential for failures. We present experimental results to support our claims.
Tuesday, August 9
Naomi Fitter, PhD Candidate, University of Pennsylvania
Advisor: Katherine J. Kuchenbecker
10:45 am, Towne 337
"Physical Human-Robot Interaction for Social Motor Coordination"
Human friends and teammates commonly connect through handshakes, high fives, fist bumps, and other forms of hand-to-hand contact. As robots enter everyday human spaces, they will have the opportunity to join in such physical interactions, but few current robots are intended to touch humans. To begin investigating this topic, we have sought to discover precisely how robots should move and react in playful hand-to-hand interactions with people.
We have conducted work in three main areas to address this design challenge. The first, machine learning from kinesthetic data, involved the observation of sensor data from people performing a variety of hand-clapping activities. Recorded accelerometer, gyroscope, and position data gathered in this phase taught us how to make a robot move in a human-like way, detect hand contact, and classify human motion intent. Developing robot behaviors was the next investigation phase, during which we used previous findings to select, modify, and program a Rethink Robotics Baxter Research Robot to play hand-clapping games with a human partner. After preliminary tests demonstrated that this robot could move like our previous hand-clapping study participants and reliably detect hand impacts through its wrist-mounted accelerometers, we performed a human-robot study to quantify how different aspects of playful human-robot interaction affect user experience. The robot's facial animation, physical reactivity, stiffness, and tempo all affected some aspects of user perception of Baxter. Several upcoming studies will help to answer additional questions about human-robot hand-clapping interactions. Finally, to broaden the scope of our designed human-Baxter interactions, we began exploring applications of Baxter in socially assistive robotics. Using many of the same sensing and actuation strategies, we developed several hand-to-hand contact-based exercise interactions to be jointly executed between a person and Baxter. In an upcoming experiment, we will test the motivational qualities of these games in human-Baxter exercise interactions.
Overall, through this work and future endeavors, we aim to help shape design processes for socially relevant physical human-robot interaction and reveal new opportunities for socially assistive robotics.
Tuesday, August 16
Tarik Tosun, PhD Candidate, University of Pennsylvania
Advisor: Mark Yim
1:30 pm, Towne 337
"Accomplishing Tasks with Modular Robots"
Modular reconfigurable robots are composed of repeated robot modules that can connect to form variable morphologies. They promise to be versatile, robust, and low cost.
SMORES-EP is a next generation, self-reconfiguring modular robot that uses electro-permanent (EP) magnets to create physical inter-module connections. This allows rapid, strong, fault-tolerant latching, as well as inductive communication between modules. We will begin with an overview of the mechanical, electrical, and software systems underlying SMORES-EP.
The advantage of modular robot systems lies in their flexibility, but this advantage can only be realized if there exists some reliable, effective way of generating configurations (shapes) and behaviors (controlling programs) appropriate for a given task. In the second portion of this talk, we present an end-to-end system for addressing tasks with modular robots, and demonstrate its capability of accomplishing challenging multi-part tasks in hardware experiments. The system consists of four tightly integrated components: (1) A high-level mission planner, (2) A large design library spanning a wide set of functionality, (3) A design and simulation tool for populating the library with new configurations and behaviors, and (4) modular robot hardware.
Finally, we present an algorithm to automatically detect kinematic and topological embedding of modular robot designs. This work expands the capabilities of our library-based framework by making behaviors portable between designs with similar sub-structures: once an embedding of morphology A in morphology B has been established, behaviors developed for A can be directly re-used with B.