Collaboratively Navigating Autonomous Systems
MQP Team: Madeline Burris, Ryan Fawthrop, Kevin Janesch, Edward Murphy, Quinn Perry
Advisors: Alexander Wyglinski (ECE), Taskin Padir (ECE/RBE)
Abstract: The objective of this project is to enable heterogeneous networks of autonomous vehicles to cooperate together on a specific task. The prototyped test bed consists of a retrofitted electric golf cart and a quadrotor designed to perform distributed information gathering to guide decision making across the entire test bed. The system prototype demonstrates several aspects of this technology and lays the groundwork for future projects in this area.
Intelligent Vision-Driven Robot For Sample Detection And Return
MQP Team: Peng Ren, Tianci Zhao
Advisor: Michael Ciaraldi (CS/RBE)
Abstract: This project explores various vision methodologies to locate and return a user-specified object. The project involves building an automatic robotic unit with an all-terrain chassis vehicle and integrated camera. The high level vision control system uses serial communication to direct the low level mechanical parts. The chosen approach for vision analysis is comparison of color thresholds. This solution provides generally accurate detection even in an environment which is noisy but has good color contrast.
Personal Assistive Robot
MQP Team: Olivia Hugal, Julien VanWambeke-Long, Kevin Burns, Nikhil Godani, Jeffrey Orszulak
Advisor: Taskin Padir (ECE/RBE)
Abstract: The aging population of the United States is creating a growing need to provide assistive care for the elderly and people with disabilities. As the Baby Boomer generation enters retirement, the ratio of caregivers to those that require assistance is projected to decrease . There are currently no commercially available modular assistive robots that can fill this need. This project aims to provide an alternative to current assistive living options through the development, construction, and testing of a Personal Assistive Robot (PARbot). This robot would allow individuals with general or age related disabilities to maintain some aspects of their independence, such as the ability to shop. Our unique solution implements a design that allows user-oriented customization. Modularity is a key component in the design to allow for future expansion and potential user customization. The robot will be designed to ADA specifications to ensure that it can operate anywhere the user desires. Human Robot Interaction (HRI) will be an important aspect in our project; users should feel comfortable in the presence of the robot. Ultimately, PARbot will be capable of navigating in public areas, such as a grocery store, and utilize a Simultaneous Localization and Mapping (SLAM) algorithm for navigation while tracking its user.
MQP Team: Patrick Bobell (RBE), Chase Cheston (ME), Ben Wilson (ME), Jacob McSweeney (RBE)
Advisor: Mustapha Fofana (ME)
Abstract: An average of 3,533 people die of drowning each year in the United States of America, over half in rivers, lakes, and beaches. For those who survive, many are left with lasting brain injury from prolonged oxygen deprivation. Distressed swimmers are rarely able to call for help or splash for attention, and often drown within 20 to 60 seconds. The primary challenge lies in immediately identifying the need for assistance and providing aid before it is too late to avoid death or permanent injury. The MQP team designed and built a Buoyant Unmanned Distress Detection and Evacuation (BUDD-E) System that can be used on crowded coastal beaches. The BUDD-E system can locate distressed swimmers and rescue them from drowning. It monitors swimmers’ pulses, identifies swimmers in distress, and communicates with lifeguards and waterfront safety staff in real time through a live display of all swimmers’ locations on a Google Earth map of the beach. In emergency or distress situations, an unmanned flotation robot is automatically dispatched at speeds of up to 20 miles per hour to victims requiring immediate support. The robot is able to transport a conscious victim to shore much more effectively than a lifeguard. The BUDD-E System compliments the efforts of rescue personnel, preventing trauma and loss of human life that would otherwise result from preventable drowning incidents.
MQP Team: Carly Buchanan
Advisors: Eduardo Torres-Jara (CS/RBE), Ryan Madan (HUA)
Abstract: The calligraphy robot is intended to act as a proof-of-concept platform for a combined control system that utilizes both position and force control with high precision. A platform consisting of a high-precision gantry system and an end-effector with series elastic actuators (SEAs) was designed and realized. PVT controllers were utilized to achieve smooth curves without noticeable shaking or jagged lines. This design has potential implications for compliant robotic applications, especially domestic and surgical incision robots.
Robot Learning Through Crowd-Based Games
MQP Team: Andy Wolff, Scott Cornman
Advisors: Sonia Chernova (CS/RBE), Rob Lindeman (CS/IMGD)
Abstract: The field of robot learning from demonstration focuses on algorithms that enable a robot to learn new abilities from examples provided by a human teacher. This project aims to show that in the context of learning from demonstration, data collection can be facilitated by collecting demonstrations from remote users through a browser and presenting the learning task as a game with certain motivational features. Furthermore, it explores possible Iimprovements on learning algorithms through the use of filtering by score, the game’s natural fitness function. We demonstrate our approach through a system that enables users to remotely control a KUKA youBot robot to play a game of whackamole through a common web browser. We collect data on how users play and utilize a decision tree learning algorithm to teach the robot to play autonomously. Using A/B testing techniques, we compare the quantity and quality of collected data between users with different motivational game features.
Indoor Navigation and Manipulation Using a Segway RMP Platform
MQP Team: Christopher Dunkers, Samuel Naseef, Brian Hetherman, Paul Monahan
Advisors: Dmitry Berenson (CS/RBE), Gregory Fischer (ME/RBE)
Abstract: The goal of this project was to work with a Segway RMP, utilizing it in an assistive-technology manner. This encompassed navigation and manipulation aspects of robotics. First, background research was conducted to develop a blueprint for the robot. The hardware, software, and configuration of the given RMP was updated, and a robotic arm was designed to extend the platform’s capabilities. The robot was programmed to accomplish semi-autonomous multi-floor navigation through the use of the navigation stack in ROS (Robot Operating System), image detection, and a user interface. The robot can navigate through the hallways of the building using the elevator to travel between floors. The robotic arm was designed to accomplish basic tasks, such as pressing a button and picking an object up off of a table. The Segway RMP is designed to be utilized and expanded upon as a robotics research platform.
Smart Robotic Prosthetic Hand
MQP Team: Sean Casley, Thanacha Choopojcharoen, Adam Jardim, Deniz Ozgoren
Advisors: Cagdas Onal (ME/RBE), Taskin Padir (ECE/RBE)
Abstract: Despite advances in upper body prosthesis industry, users of this technology still struggle to complete everyday tasks with the same level of ease as they once could. Complaints of available products continue to be high market costs as well as complex user interfaces, limiting the effectiveness of the device. As a solution to these issues we propose the creation of a Smart Prosthetic Hand – a less-expensive semiautonomous robotic prosthesis capable of determining the most appropriate grip for grasping an object then executing that grip. The device we propose is an anthropomorphic prosthesis with independently movable fingers capable of executing a variety of grips. Autonomy is achieved through the use of object recognition and closed-loop control system which will take input from a camera in the palm to determine the shape of an object, the position of each finger, and quality of grip. Then the identified grasp will be executed based on the user's input and controlled by using hybrid position/force control algorithm.
MQP Team: Jillian Chalke, Greg Hutchinson, Paul O’Brien, Victor Puksta, Christopher Conley
Advisor: Mustapha Fofana (ME)
Abstract: The use of professional scuba diving teams is an industry practice for multiple commercial, research and military applications. Professional divers and be deployed into dangerous conditions where surface communication can be limited. In these situations, remotely operated vehicles (ROV’s) can be implemented to improve communication and ensure personnel safety and mission success. The Dangerous Inspection and Versatile Exploration Robot (DIVER) is design to assist, track and monitor professional divers in commercial, military and research applications. The remotely operated vehicle is designed to be lightweight and versatile, allowing for rapid deployment amongst professional scuba diving teams. Using a combination of commercially available and custom made components, the DIVER team created a user friendly and highly mobile ROV platform. The team conducted background research, determining specific applications and features necessary for the DIVER ROV. Mechanical engineers designed components for manufacturability, strength and compactness. Robotic engineers integrated electrical and software components to allow for usability and functionality. Advanced real time tracking algorithms allow for autonomous operation while integration of a video game controller provides the user with a familiar platform for tele-operational driving. The overall design allows for platform expandability, providing the user with options for upgrading individual components or creating specialized payloads based on specific applications.
Intelligent Surveillance UAV
MQP Team: Arianna Niro, Andrew Gallagher, Ben McIntyre, Steven Guayaquil, Antonio Puzzi, Arman Uygur
Advisors: Taskin Padir (ECE/RBE)
Abstract: Surveillance is critical for military, law enforcement, and search and rescue operations. In the past, stealth aircraft and helicopters were used for these types of missions. Recently however, unmanned aerial vehicles (UAVs) have grown in popularity and are an excellent resource that can be utilized for surveillance missions. Since there are many drones capable of this, this project sought to create a surveillance UAV that was autonomous, inexpensive, lightweight, and easy to manufacture. The drone was designed as a quadrotor that houses two cameras and a wireless transmission system that provides live feed from the cameras to the ground station. It was also designed to be able to carry a payload for future developments.
Enabling Autonomous Manipulation on iRobot’s PackBot
MQP Team: Alex Henning, Nick Morin, Jess Gwozdz, Ransom Mowris
Advisor: Dmitry Berenson (CS/RBE)
Abstract: iRobot's PackBot is the world's most successful defense and security robot. It has been deployed at Ground Zero, the Boston Marathon, Fukushima, Iraq, and countless other scenarios. Currently, PackBot is entirely tele-operated, meaning a user needs direct every motion on the robot. This requirement compromises the situational awareness of operators and might put them in danger. Our project solves this problem by providing a platform for semi-autonomous manipulation by means of the Open Robotics Automation and Visualization Environment (openRAVE). We demonstrate the expandability of our platform by implementing "click-to-grasp" functionality for objects recognized by a 3D camera.
Automated Tool Prep
MQP Team: Ryan Wheeler, Eric Willcox, Sergey Zolotykh
Advisors: Craig Putnam (RBE), Stephen Nestinger (ME/RBE), Lifeng Lai (ECE)
Abstract: This MQP has two goals that aim to show the feasibility of automating some tasks for General Electric Aviation. The first goal was to design a system capable of changing used end mills within a tool holder used by CNC machines. The second was to create a method to automatically keep track of the location of all tool holders as well as keep a log of all jobs on which they have been used. The end results showed that both are possible and with more work could be implemented in a working system.
Test Bed for Flapping Wing Robotics
MQP Team: Jesus Chung, Chris Overton, Kevin Ramirez, Tyler Pietri, Alexandra Beando
Advisors: Marko Popovic (PH/RBE/BME), Stephen Nestinger (ME/RBE)
Abstract: Ornithopters, bio-inspired systems that utilize flapping wing flight to generate lift, are a growing field of robotics with a wide range of applications. Although these bio-inspired robots are of particular interest, there are currently no successful large scale hovering ornithopters over 2kg in existence. Continuing from last year’s MQP, this project developed a test bed that can effectively examine ornithopter designs and further flapping robotics research. The realization of the test bed was guided by a theoretical model developed in MATLAB. Utilizing load cells, cameras, and a LabVIEW interface, the test bed allows for the examination of different wing designs and wing motion.
MQP Team: Matthew Simpson, Emanuel Demaio, David Ilacqua, Louie Mistretta
Advisors: Michael Gennert (CS/RBE), Ken Stafford (ME/RBE)
Abstract: Project Squirrel is a response to the need for a robotic mechanism with the capacity to traverse vertically oriented trees and gather information. The growing Asian Long-horned beetle infestations in the United States are one of the most prominent producers of this need. Current solutions to this issue are largely inefficient and could greatly be improved through the implementation of robotic mechanisms. The design and production of a robotic device which could be deployed by a single operator and used to ascend and descend trees of varying dimensions would greatly reduce the time and labor needed to address these infestations. Project Squirrel primarily seeks to address this need through the design and production of a robot which improves current processes used to monitor Asian Long-horned beetle populations. Project Squirrel has built upon past ideas to create a tree climbing robot and much of its design has been inspired by the successes and failures of other robots. Background research in other projects and solutions has aided in steering the direction of project design. As a result of extensive research, testing, and manufacturing, team has successfully been able to design and build a robot both mechanically and electrically from the ground up.
Exomuscular Sleeve for Upper Limb Stroke Rehabilitation
MQP Team: Michael Cross, Tim Forrest, Jason Klein, Anselm Mak, Julieth Ochoa-Canizares, Hosung Im, Ritesh Adhikari
Advisors: Gregory Fischer (ME/RBE), Cagdas Onal (ME/RBE), Ted Clancy (ECE)
Abstract: Traditional physical therapy for upper-limb post-stroke hemiparetic patients suffer from time and money complications, often fails to reach the maximum potential for recovery, and is unable to provide a complete, quantitative assessment of a patient’s progress. Through the use of robotics, the team aimed to create a device which would not have these shortcomings and would provide a holistic physical therapy solution. The sleeve achieves exomuscular actuation through a system of Bowden cables linked to DC motors housed remotely and is able to flex and extend the fingers and elbow and control pronation and supination of the wrist. Through a sensor array located throughout, a feedback system is able to collect quantitative data on position and pressure, and control all degrees of freedom utilizing these data and several on-board processors.
Optimal Driveline Robot Base
MQP Team: Stephen Diamond, William Dunn, Kirk Grimsley, Michael Cullen
Advisors: Taskin Padir (ECE/RBE), Ken Stafford (ME/RBE)
Abstract: Our team has decided that there is currently a need for a driveline system that is capable of performing a zero radius turn and being maneuverable at low speeds while also maintaining traction, stability, and energy efficiency at high speeds. We designed and prototyped a modified Ackermann steering system driven by a single motor, with an extended range of motion. The steering system was integrated into a robot chassis that meets FIRST Robotics Competition requirements.
Sailboat Stabilization System
MQP Team: Mike Brendlinger, Dominic Gonzales, Pedro Miguel
Advisors: Ken Stafford (ME/RBE), Bill Michalson (ECE/RBE)
Abstract: Sailboats have played an integral part in history and the development of modern society. The scope of this project focuses on one particular aspect of the operation of this boat: the crew. The crew is the secondary boat operator sitting toward the front of the boat and the primary need for the crew is to simply shift his/her weight in the boat in order to keep the boat at the desired angle. By replicating this action with an autonomous device, the crew can be eliminated. By setting a desired heel angle on a control panel, a mass driven by a motor can be moved laterally along a track in order to adjust the heel angle accordingly. The goal of this project is to allow a sailor to safely operate a Flying Junior dinghy without a crew and be able to maintain full control of the boat.
Development of Multi-modal Control Interfaces for a Semi-Autonomous Wheelchair
MQP Team: Ran Li, Gilmar da Vitoria, Runzi Gao
Advisor: Taskin Padir (ECE/RBE)
Abstract: The purpose of the project is to assist users with different levels of disabilities to control a semi-autonomous wheelchair. A semi-autonomous wheelchair developed by RIVeR Lab is able to perform assistive control to avoid obstacles and cliffs and to follow walls. With a joystick control adapter, the basic joystick of the wheelchair can take commands directly from computers. In addition to joystick mechanical adapter control, human-machine interaction and control methods such as voice and electromyography (EMG) are deployed, with the aim of enabling people with different levels and types of disabilities to control the wheelchair. These non-physical motion based user control interfaces allow people with limited mobility to control the wheelchair with a desired accuracy.
MQP Team: Gregory McCarthy, Nicholas Corso, Brian Jennings, Daniil Effraimidis
Advisors: Marko Popovic (PH/RBE/BME), Cagdas Onal (ME/RBE)
Abstract: The Exo-Musculature project is a novel hydraulically-actuated elastic muscle, which is inspired by the capabilities of a biological muscle. The design has certain advantages over natural muscles, such as being able to maintain a position without expending energy. This design uses an elastic element to apply tensile force, which is released when hydraulic pressure is applied. This gives the muscle the unique characteristic of storing elastic energy when pressurized and releasing it to contract. Other artificial muscles, such as the McKibben,are similar to ours in the respect that they are fluid-actuated and can be locked in place, but the McKibben requires suction or pressure to expand and contract. Additionally, our muscle is limited to expansion in only one dimension, which offers a higher energy density. It is much more compliant than traditional hydraulic cylinders, making it better suited for use in human rehabilitation and augmentation. Finally, our prototype is constructed using common materials, making it an extremely low cost solution for both medical and robotic applications.
Experimental Validation of a Scalable Mobile Robot for Traversing Ferrous Pipelines
MQP Team: Sarah Sawatzki, Frederick Baruffi, Brad Mello, Taylor McNally
Advisor: Taskin Padir (ECE/RBE)
Abstract: This project involved the design, construction, and testing of magnetic wheels for use on a mobile robotic platform developed for the laser welding of oil and gas pipelines. The current process is a dull, dirty, and dangerous job that can easily be delegated to an enhanced process with the use of a mobile robot. The goal of the project was to experimentally validate a scalable magnetic wheel design for use on a mobile robot platform. This involved researching and testing various types and configurations of magnets for use in magnetic wheels, constructing test fixtures, modifying a test chassis, as well as developing a control algorithm to accurately track the weld seam and traverse the pipe.