Final Project

Project Description
Student Project Pages


Brief Description

Your final project grade will be determined by:

  • final presentation
  • final report (text + video)

Final Presentation

When: Monday, May 2 at 11:00am-2:00pm
Where: Levine 307

  • 5 minutes per person
  • 2 minutes Q/A from the Professors & TAs
  • Q/A from fellow classmates through wiki or googlegroups that the person needs to address in the report)

Your presentation should cover:

  • problem statement
  • what you have done
    • video of simulation or experiment
  • what remains to be done in one week (feel free to complete the project by the presentation deadline)

If your project has not been finished by Monday, May 2nd, your presentation will be scheduled for a later date. Email a preliminary report to us on Friday, April 29, showing what has been completed (> 90%) and what will be completed over the weekend so we can schedule you for the presentation.

Final Report & Video

Due: Monday, May 16 at noon

FAQ on editing the Wiki

Project Pages

Robot Coordination with Underactuated Robots
Anirudha Majumdar
Underactuated robots can often exploit their natural dynamics in order to achieve better energetic performance than robots that are fully actuated. Passive dynamic walkers are an extreme example of this. However, the lack of complete control authority that results from underactuation makes robot coordination tasks much more difficult. The goal of this project is to explore coordination tasks with underactuated robots. In particular, the task that will be considered is the passing of a soccer ball back and forth between two underactuated compass gait walkers.

Implicit C-space Constraints Imposed by Joint Torque Limits
Avik De
The goal of this project is to investigate the obstacles in configuration space imposed by constraints on joint torques for a 3-link planar manipulator. Large end-effector loads may result in large joint torques depending on the joint configuration, and constraints on the former result in regions of C-space being rendered impassable (for known, fixed load). A secondary goal is to find paths through configuration space (for the manipulator with payload) that are "optimal" in terms of path length or energy consumption, while keeping these constraints in mind.

Get the door! An Example of Cooperative Robotics
Ben Charrow
My project will focus on coordinating a heterogenous pair of robots: the PR2 and the Scarab. While the PR2 has a large footprint and is equipped with a wide array of sensors and actuators, the scarab is small differential-drive robot equipped with a single laser scanner. Despite their differences in complexity, an autonomous team would benefit from having both robots as members. As a demonstration of how these two robots can work together, I plan on having the PR2 open a door for the scarab, allowing the scarab to explore a new environment.

Tracking and Chasing Via Attached Camera
Brandon Fischer
The police sometimes must apprehend a fleeing suspect and thus must pursue until one of the parties slips up where either an escape or an arrest is made. Automated police vehicles would need to use a visual tracking system to watch the pathing of the suspect in order to follow. This project will involve working with methods of tracking a given objects through a camera attached to the vehicle with depth information in order to avoid obstacles and attempt to maintain visual contact with the target, sometimes requiring maneuvering around an obstacle to regain that vision.

Multi-dimensional Planning for Autonomous UAVs
Brian MacAllister
Path planning in large environments can be expensive both computationally and memory wise in higher dimensional space. One approach to this problem is to reduce the dimensionality of the planning environment in some areas while keeping the original dimensionality in others. The goal of this project focuses on the development of a multi-dimensional planner for use in planning trajectories for autonomous unmanned aerial vehicles, more specifically rotorcraft, in a simulated environment. It should be capable of re-planning trajectories within a reasonable amount of time.

Adaptive Control for a Mechanical Manipulator
Caitlin Powers
One of the key challenges facing robotics is the development of reliable autonomous manipulation. The goal of this project is to develop a simulation of an adaptive controller applied to a 4 DOF manipulator, modeled on the Barret WAM, as it lifts an object and carries it through a trajectory. The controller will estimate the mass and inertial properties of the object as it is hefted. The question of when the properties of an object can be reliably estimate will be addressed.

Trajectory Planning for Multi-hop Communications with a Probabilistic Radio Fading Model
Cem Karan
This project will concentrate on planning trajectories for groups of robots in a known environment, but with communications that have only probabilistic guarantees for connectivity, and which are dependent on the environment the robots are traveling through. The robots will need to maintain communications to a base station by routing packets from one robot to the next in a chain, and will need to move to permit the base station to control a lead robot, regardless of where the lead robot is in the environment. The intermediate robots will need to move as needed to ensure that the lead robot will always have communications to the base station.

Micro-robotic Path Planning and Control for Single Cell Manipulation
Ceridwen Magee
Single cell manipulation through the use micron-sized chemical beads and micro-robots is an ongoing collaboration between the GRASP and biology labs. This project focuses on creating a user-friendly interface powered by a simple A-star planner for targeted delivery of chemicals. The general goals will be similar to MEAM 620 project one, but backed by actual implementation in an experimental setting. Working in collaboration with Ed Steager.

Contemporaneous Motion and Communication Control in a Robotic Network
Chinwendu Enyioha
A mobile, wireless network with a task (for example, real-time sensing and communication over an area), is considered. Of interest is the problem of determining optimal motion paths (for agents under tight communication constraints), which yield high network throughput while keeping the powers expended for transmission low. We want to understand the analytical or computational relationship that exists between agent positions and network routing and flow rates. With this study, it is also possible that the trade-offs between max throughput and min power can be exploited. The communication model being considered here has a fading component, on which the transmitted powers and network flow rates depend.

Multi-robot Path Planning in 4D Configuration Space
Fei Miao
This project attempts to solve path planning and multiple robot coordination problems. A* star search is applied for two robots finding their collision free shortest paths from initial configurations to reach their final configurations. When there are more than one robot in the world space, the robots need to know how to correspond to each other so that collision is avoided and the state space is 4D in this problem. There will be a general comparison between another algorithm that is designed for efficiently searching high-dimensional spaces, named Rapidly-exploring Random Tree (RRT). The second method is based on a RRT data structure and also has widely application, although the solutions found may not be the optimal solutions.

Magnetic Structure Construction with the PR2 Robot
Ian McMahon
This project will utilize the PR2 robot to explore autonomous construction of magnetic blocks. In doing so, it will be necessary to assess the error inherent in detection of the building blocks and their orientation, and how that error affects the subsequent manipulation of two block pieces. Furthermore, an impedance controller will be utilized in order to allow the robot to passively connect the magnetic pieces into a locking position. This project will incorporate resources from the open source ROS community.

Backpack Detection using Image and Depth
Jason Owens
All robots that aspire to be autonomous need to detect and recognize objects of various kinds. This project will focus on the detection and classification of backpacks and similar classes of deformable bags in an initially constrained environment. Since it is not uncommon for explosives, weapons and other contraband to be concealed inside duffel bags or backpacks, this capability could be especially useful in a variety of safety or security sensitive scenarios. We will explore various feature detectors and learning algorithms with the goal of designing a perceptual component that provides high-confidence detections of bags in various configurations along with pose estimates of salient features.

Coverage of an Area with Multiple UAVs
Kartik Mohta

Given the probability distribution of an event happening in an area, a groups of UAVs (quad-rotors) would be deployed to cover the area so as to maximize the probability of detection of the event. The optimal placement of the nodes in such a scenario are the generator points (the centroids) of a Centroidal Voronoi Tessellation (CVT) of the region. Thus the project involves calculating the CVT of a region based on the given probability distribution and then planning trajectories for each UAV such that finally they reach the desired configuration (i.e. a UAV at each centroid of the CVT).

Literate PR2
Menglong Zhu
The ability to detect and read text in natural scenes is important for a robot to explore, localize and navigate autonomously in unknown environments. Comprehending text provides meaningful context for a robot to understand its surroundings and can be really useful for household robots to recognize labeled objects. Apart from understanding the scene, the ability to speak provides interactivity between robot and the real world, which can also be useful in indoor settings. The goal of this project is to apply state-of-art text detection and recognition algorithm to PR2 robot and achieve the task of reading out printed text and signs in indoor environments. The deliverables will be a video demonstrating the robustness of the system and a releasable software package that is integrated with ROS.

System for Target Location and Immobilization
Monica Lui
The goal of this project is to design a system of hunter mobile robots set to locate and capture an elusive target. Each hunter robot will sense and communicate its estimate of the target's location to neighboring robots in order to determine the location of the robot, with some uncertainty. The robots will attempt to surround, or immobilize, the target.

Multi-robot, multi-target tracking
Philip Dames
My project is to control a group of robots to explore an unknown environment and track any objects of interest that they spot along the way. I will use a group of several scarab robots equipped with laser range finders to estimate target locations, relay these estimates to their neighboring robots, and follow the targets. However, the control scheme must balance the goal of tracking targets that have been seen with keeping the scarabs spaced out in the environment to increase the chances of spotting new targets.

Torque control of a Throwing Trajectory by a WAM
Shuai Li
This project will focus on torque control of WAM and trying to find the max speed of the end effector of the arm. Before testing it on a WAM, dynamic simulation will be used to get the trajectory and speed of the arm in Matlab. The dynamic simulation will use Lagrange equations to solve the dynamic module. Then use Matlab again to generate the torque timetable using PID control and finally, the goal will be to test it on a WAM.

Collaborative Mobile Manipulation Among Humanoid Robots
Steve McGill
Enabling mobile robots to collaborate on tasks exploits synergies that no independent robotic actor could achieve. However, implementing the required robot-robot interactions induces extra constraints on the individual robot controllers. Presented will be a cooperative system where two DARwIn-OP humanoid robots pick up a table and carry it to a given destination. The induced constraints include synchronization of bipedal walking gaits, coherent object state modeling and coordinated mobile manipulation.

Path Planning and Control of a 4-Legged Robot using an Android-based Device to enable Real Time Mobile Tele-conferencing and Telepresence
Teyvonia Thomas
Recent advances in mobile communication devices and cloud based technologies have expanded the realm of possibilities for personal robotic applications. This project will integrate some of these technologies with key components of MEAM 620. The objective is to plan optimal trajectories for a 4 legged articulated robot (AnguKo) in a known environment with obstacles. Planning will be done using the A Star algorithm. Color coded markers and blobs will aid in obstacle detection and avoidance. The robot will be remotely controlled using an Android based mobile device. Successful implementation of the motion planning and vision-based algorithms will enable AnguKo to navigate an indoor environment without collisions and therefore operate more efficiently as a mobile tele-conferencing and telepresence robot.

Real-Time Obstacle Detection for Humanoid Robot with Single Camera
Yida Zhang
Obstacle Detection is essential for Humanoid Robot on both localization and motion planning. However, due to the Single RGB Camera and limited computation capacity, real-time obstacle detection remains a challenge. This project will focus on enabling robots to detect color marked obstacles and find a optimal path in real time for specified target. The major constraints include shadow recognition, obstacle state modelling and positioning. Final demonstration will be hold on DARwInOP humanoid robot. Given a target, Robot will automatically walk toward the target throught an unknown area with obstacles.