Research

From RoMeLa

Jump to: navigation, search

Contents


Current Research

PIRE: Humanoids-University Accessible Infrastructures to Advance Capabilities

In partnership with researchers from Drexel, UPenn, Swarthmore, Bryn Mawr College, and international partner Korea Advanced Institute of Science and Technology (KAIST) this project will bring leading roboticists from the US and Korea together to advance state-of-the-art humanoid robotics. The project will result in infrastructures that will produce far-reaching broader impacts and will enable humanoids to work and socially interact with people. Virtual-Hubo, Mini-Hubo, and a remotely operable Hubo test rig are being developed. The budget for this five-year project is $2.5M.

Researchers: Jeakweon Han, Karl Muecke

DARwIn: An Analytical Motion Filter for Humanoid Robots

The Analytical Motion Filter (AMF) takes reference motion of humanoid, and stabilizes an otherwise unstable motion while preserving as much of the reference motion as possible. The analytical solutions will provide insight to the stability of the robotic system that would otherwise be difficult to identify in a numeric optimization scheme. The specific objectives are to: stabilize reference motions while retaining as many of the characteristics as possible; verify the filter using human motion capture data, kinematically synthesized data, and motion data from animation software in both simulation and on a humanoid robot platform; and provide insight into the stability of the humanoid robotic system.

Researcher: Karl Muecke

DARwIn: Precision Circular Walking of Biped Robots

Whenever bipedal robots need to make turns, the ability to walk stably and precisely along an abitrary radius curve will be quite beneficial. This motivates us to derive new ZMP constraint equations with respect to a rotating coordinate frame, seek appropriate dynamic gaits based on them, and address the forward and inverse kinematics. A set of dynamic walking patterns including the transient are are herein proposed and applied to an exemplicative case of turning locomotion. Dynamic simulations prove the patterns to be successful even in the presence of distributed-mass and ground contact effects, and experiments using the DARwIn humanoid robot platform will be conducted.

Researchers: Dr. Seungchul Lim, Karl Muecke

DARwIn's Brain-A.I. for a Soccer Playing Humanoid Robot

The goal of this project is to provide embodied Artificial Intelligence to a multi-agent team of humanoid robot DARwIns such that they can compete in the International RoboCup competition. RoboCup is a landmark project in the robotics and A.I. communities, presenting a standard problem of developing a team of fully autonomous humanoid robot soccer players that can beat the human world soccer champions by the year 2050. This project is focused on building a robotic control architecture capable of performing vision processing, real-time sensor fusion, high-level behavioral management, cooperative teamwork, and bi-pedal gait generation for dynamically stable walking.

Researcher: Jesse Hurdus

DARwIn: Stabilized Omni-directional Walking Engine for Miniature Humanoid Robots

The goal in this research is to implement omnidirectional and stabilized locomotion. Omnidirectional locomotion is the ability to move in any direction regardless of the orientation of the robot. This is a huge advantage in dynamic environments and restricted spaces such as robot soccer. We will use the body to track the on-line ZMP reference to stabilize walking. All the high-level programming will be written in MATLAB and the lower-level interface to the hardware will be implemented by C++. Webots will be used for simulation purposes in this research.

Researcher: Seungmoon Song

Development of an Autonomous Vehicle for the DARPA Urban Challenge

'Odin' is Team VictorTango's entry in the 2007 DARPA Urban Challenge, an autonomous ground vehicle competition in an urban environment. The team includes 46 undergraduate students, 8 graduate students, 4 faculty members,5 full time TORC employees and industry partners, including Ford Motor Co. and Caterpillar, Inc. Team VictorTango successfully completed the DARPA Urban Challenge final event, finishing 3rd with a cash prize of $500,000. During the competition, Odin was able to drive several hours without human intervention, negotiating stop sign intersections, merging into and across traffic, parking, and maintaining road speeds.

Researchers: Jesse Hurdus, Shawn Kimmel, Team VictorTango

Blind Driver Challenge: Developing Non-visual Driver Interfaces

The goal of The Blind Driver Challenge (BDC) is to improve the mobility and independence of visually impaired persons through advancements in robotics technology. We have developed vehicle perception, information processing, and non-visual driver interfaces, including a novel audio and tactile feedback system which conveys both heading and speed information to the driver. This system has been benchmarked by visually impaired persons through an affiliation with the Virginia School for the Deaf and the Blind. The BDC has been closely tied to the DARPA Urban Challenge project at Virginia Tech, and the award winning autonomous vehicle "Odin" may be used for the base vehicle platform for the deployment of the BDC system.

Researchers: Shawn Kimmel, Team BDC

CAREER: Whole Skin Locomotion: Structural Design and Actuator Development

For this project, we apply fundamental analytical and experimental research to the design and development of the Whole Skin Locomotion (WSL) platform. The specific objectives are to: develop a stability/buckling model for hollow tube structures loaded in flexure and internal pressure; develop a quasi-static model for the fluid filled WSL platform under circumferential ring actuator force; develop and simulate a dynamic model of the WSL mechanism and conduct a sensitivity analysis for actuator parameters; verify performance characteristics through a prototype that implements design parameters and control.

Researcher: Eric Williams

CAREER: Whole Skin Locomotion: Development of an Incrementally Loaded FE Model

The goal of this research project is to develop a model to which we can apply actuator forces and predict the resulting shape and motion of the unique Whole Skin Locomotion (WSL) robot. The objectives are to: gain an understanding of finite element models used for membrane surfaces and implement a suitable model with an appropriate loading strategy; find the final geometric shape of the robot and predict its motion given an environment and actuator loads; develop an analytical solution and compare it to the FEA results to gain insight into the fundamental mechanisms that propel this robot.

Researcher: Derek Lahr

IMPASS: 3D Kinematic Analysis Based on Screw Theory

IMPASS is a wheel-leg hybrid robot that can walk in unstructured environments by independently extending, or retracting, three actuated spokes of each wheel. The current research objectives for this project are: classification for topology structures of IMPASS based on different ground contact points; mobility analysis for different configuration cases, using both conventional and screw-based modified GrĂ¼bler and Kutzbach criterion; inverse and forward position analysis for the critical topology scheme of IMPASS; singularity configuration identify and investigation using screw theory; and screw-based Jacobian analysis.

Researchers: Ya Wang, Ping Ren

IMPASS: Reactive and Deliberative Motion Control for Rough Terrain Locomotion

This project researches motion planning strategies for the novel actuated spoke wheel robot, IMPASS. The specific objectives are to: develop 2-D and 3-D motion planning strategies in unstructured terrain for both terrain sensing and non-terrain sensing configurations; verify motion planning strategies in simulation and experimentally; advance the capabilities of the hardware platform, including a moving center of gravity, onboard computer and power, and rugged body and components; develop accurate and dependable perception units for terrain sensing and object recognition, including laser range finders and cameras.

Researchers: Shawn Kimmel, Blake Jeans

STriDER: Gait Planning and Standing Up Strategies

This research investigates standing up strategies for the novel three-legged robot STriDER. The unique structure and operation of STriDER makes the simple task of standing up challenging: the relative height and long limbs of the robot require high torque from the actuators due to large moment arms; the joint configuration and length of the limbs limit the workspace where the feet can be placed; the compact design of the joints allows for limited actuator torque; the number of limbs does not allow extra support and stability in the process of standing up. Five strategies have been studied: a three feet pushup, two feet pushup, one foot pushup, spiral pushup and feet slipping pushup.

Researcher: Ivette Morazzani

STriDER: Kinematic Analysis of its Parallel Configuration

STriDER (Self-excited Tripedal Dynamic Experimental Robot) is a unique walking robot with three legs. When not walking, STriDER can be modeled as a three-branch in-parallel manipulator given the assumption that all three foot contact points are fixed on the ground with no slipping. The conclusions derived from the kinematic analysis will be utilized in dynamic analysis and motion planning. The specific research objectives are: solve the inverse and forward displacement problems, establish the Jacobian matrices, identify the singularity and propose the elimination method based on redundant actuation.

Researcher: Ping Ren

CLIMBeR: Cable-suspended Limbed Intelligent Matching Behavior Robot

CLIMBeR (Cable-suspended Limbed Intelligent Matching Behavior Robot) is a robot being developed for climbing unstructured cliffs. Utilizing a multi-contact force distribution algorithm and by adjusting its posture for stability. CLIMBeR uses matching behavior (swapping foot to foot, hand to foot, or hand to hand on a single hold or point) to plan each foot placement. CLIMBeR has three 3DOF limbs, a winch with a cable and will soon have a miniature laser range-finder to sense the terrain geometry for planning. This project is sponsored by NSF as a REU project.

Researcher: Brad Pullins

CHARLI: Cognitive Humanoid Autonomous Robot with Learning Intelligence

CHARLI (Cognitive Humanoid Autonomous Robot with Learning Intelligence) is a 1.3m tall, autonomous humanoid robot, which will be the first of its kind in the United States. Partially funded by Virginia Tech's SEC, and sponsored by National Instruments, Maxon, and Robotis, this humanoid robot will be used as a platform for research, education, outreach, and publicity for Virginia Tech. The completed humanoid robot will be able to autonomously navigate the hallways of campus buildings walking on two feet, and perform human like complex motions such as giving tours indoors.

Researchers: Jeakweon Han, Team CHARLI

Hardware-Accelerated Nonlinear Predictive Control for Legged Locomotion

We plan to realize adaptive and resilient locomotion in legged robots by implementing a novel hybrid hard-accelerated nonlinear predictive control architecture inspired by biological nervous systems. This effort will be a first step in a larger collaborative research effort in studying the hierarchical relationship found in the nervous system of animals for locomotion control; in particular, to understand how higher-level centers' predictive capabilities can modulate lower-level centers' gait generation, and to translate this knowledge to engineering design principles for adaptive and resilient gait generation. The 1.3 tall humanoid robot CHARLI will be used to test the control system.

Researcher: Joe Hays

CIRCA: Climbing Inspection Robot with Compressed Air

CIRCA (Climbing Inspection Robot with Compressed Air) is a unique climbing robot that utilizes McKibben air muscles to climb scaffolding structures for inspection tasks. The prototype design won First Place in the CAGI (Compressed Air and Gas Institute) Innovation Awards design competition in 2008 with a cash prize of $10,500. The current focus of the research project is the analysis of the continuum mechanics and motion generation algorithms, while the full size prototype is being developed by a group of senior undergraduate design project students.

Researcher: Nick Thayer

HyDRAS-Arm: Automatic Calibration and Intuitive Control of Manipulator Arms

With partnership with OpenTech, Inc. (Manipulator SBIR Phase One) HyDRAS-Arm (Hyper-redundant Discrete Robotic Articulated Serpentine-Arm) is a 9+ DOF serpentine manipulator arm controlled with a full 3D shape sending 'shape tape' for teleoperation. Automatic Calibration and Intuitive Control of Manipulator Arms (AIMs) is a software system that provides simulation, visualization, and advanced control of robotics manipulator arms using real-time genetic algorithms.

Researcher: Mark Showalter

HyDRAS-Ascent: Design and Analysis for a Pole Climbing Serpentine Robot

By using a series of actuated universal joints in a helical configuration, HyDRAS (Hyper-redundant Discrete Robotic Articulated Serpentine) can wrap around a pole and use a relative neutating motion between its modules to propel itself along a pole structure. The current focus of the research is to find the relationships between design and operational parameters for optimization. Additionally, the force and torque analysis is being completed. A full scale prototype robot is also being developed by a group of senior undergraduate student design team.

Researcher: Gabriel Goldman

CIVT: Cam-based Infinitely Variable Transmission

The Cam-Based Infinitely Variable Transmission (CIVT) is a novel, highly configurable, ratcheting infinitely variable transmission (IVT) utilizing a three-dimensional camoid (patent pending) based on the operation of a planetary gearset. It is unique in both its operation and its possible applications. It combines the flexibility of both a planetary gearset and an IVT into one package. Unlike other ratcheting IVTs which produce a nonuniform output for a uniform input, this transmission can shape the output to match many periodic waveforms. Therefore, this ratcheting drive has the unique ability to produce a uniform and continuous output.

Researcher: Derek Lahr

ReCoM: Revolute Compliant Mechanism-Design Methodology

ReCoM is a novel compliant revolute joint suitable for both micro and macro scale applications (patent pending). This device is a simple, monolithic, and planar mechanism which incorporates a number of interconnected flexible spokes, radially positioned between a hub and rim. Because of the unique linkage system connecting the spokes, they flex when a torsional load is applied to the hub but remain stiff to radial loads. Compliant mechanisms also have zero backlash, are low cost, and do not wear, making them ideal for use in harsh environments and as replacements for the conventional mechanical systems in MEMS.

Researcher: Derek Lahr

MARS: Workspace Analysis and Hexapod Gait Generation

The Multi-Appendage Robotic System (MARS) is a hexapedal robotic platform inspired by JPL's LEMUR IIa robot. Each of the six limbs of MARS incorporates a 3DOF, kinematically spherical proximal joint, and a 1DOF distal joint. The generation of walking gaits for such robots with multiple limbs requires a thorough understanding of the kinematics of the limbs, including their workspace. In this research we develop the workspaces for the limb of MARS in the knee up configuration which range from simple 2D geometry to complex 3D volume, and analyze its limitations for use in walking on flat level surfaces, and apply it to the development of adaptive walking gait algorithms.

Researcher: Mark Showalter

MARS: Gait Generation with CPG for Unstructured Terrain

Researcher: Robert Mayo

Inertially-Actuated Passive Dynamic Step Climbing Wheeled Robot

A novel wheeled robot with an inertially actuated sliding spring-mass system to increase the mobility without requiring active actuated is developed. Accelerations and decelerations of the robot causes the sliding mass to shift forward or backward, accomplishing both the lifting and landing actions necessary to propel a robot over a step. The effective center of gravity of the robot will change as the sliding mass moves and a spring reaction will add an additional torque on the robot. If designed properly, these effects can allow the robot to lift its wheels off the ground - one axle at a time - and hop over the step.

Researcher: John Humphreys

A Portable Approach to Behavioral Programming for Complex Autonomous Robot Applications

In this research, an approach to behavioral programming is developed that provides the designer with an intuitive method for building contextual intelligence while preserving the qualities of emergent behavior present in traditional behavior-based programming. This is done by using a modified hierarchical state machine for behavior arbitration in sequence with a command fusion mechanism for cooperative and competitive control. this presented approach is analyzed with respect to portability across platforms, missions, and functional requirements. Specifically, two landmark case-studies, the DARPA Urban Challenge and the International RoboCup Competition are examined.

Researcher: Jesse Hurdus

Completed Research

Design of a Novel Joint Aligning Mechanism for Legged Robots

Researcher: Derek Lahr

Mobility of Autonomous Vehicles in Costal Terrain

Researcher: Ping Ren

Experimental Study on the Mobility of Lightweight Vehicles on Sand

Researcher: Marilyn Worley

Whole Skin Locomotion: Mechanics of the Concentric Solid Tube Model

Researcher: Mark Ingram

Design of a Novel Tripedal Locomotion Robot and Simulation of a Dynamic Gait for a Single Step

Researcher: Jeremy Heaston

IMPASS: Kinematic Analysis in the 2D Sagital Plane

Researcher: Doug Laney

Views
Personal tools