whuang@cs.rpi.edu
Brian Kimball, Sam Hopewell, Eugene Mulone, and Chris CernigliaWe plan to model human manipulation of rhythm sticks and apply it to a robotic manipulator. We plan to begin with a one degree of freedom robot that will use a U shaped end effecter connected to a motor. The end effecter will pivot back and forth keeping the rhythm stick in a state of equilibrium. Initially, the motion of the system can be constricted to two dimensions simplifying the complexity of the task.
Tristan Shone, Greg Krudysz, and Keith BrownOur project will involve the basic cup in ball (kendama) project, using a 1 degree of freedom arm, where we will place a cup at one end and tie a string with a ball attached at the other end. The objective is to plan an arm motion and design an algorithm to control a motor that swings the ball upso that it lands in the cup.
Ritesh Pradhan and Melaku AyeleTo design a system which identifies the configuration of the object on a "Pachinko" machine.
We are planning to append the existing design of the "Pachinko" machine on which Prof.Akella had worked in Urbana Champagne . As per our discussions with Prof.Akella, we will be working on the sensory aspect of Pachinko machine.
The Pachinko machine, at present, orients the objects from the assumed initial configurations to the specific goal configuration using of the array of pins. These pins are actuated by means of a program .
A drawback of the present method of manipulation is that there exists a lack of initial configuration identification. We are planning to introduce a technique to determine the initial configurations.
Kyle Baldassari, Justin Gullotta, Gordon Mac Millan, and Matt MangerMake a yoyo go up and down in a periodic motion
Joshua Hort, Daniel Burkott, James Woodson, and Ian BoydWe propose to implement the cup in ball (kendama) project. By using a 1 degree of freedom arm, we will place a cup at one end and tie a string with a ball attached at the end. Our objective is to plan an arm motion that swings the ball up so that it lands in the cup.
Derek Dalrymple, David Bechberger, Allen Chein, and Xianfeng ZhaoOur project is to use a mobile robot to first map a unknown configuration space. The robot will also be placed in an unknown location and it will be able to locate itself from the map it has previously made. After it has discovered its location it will plan the most effective path to a goal and follow it.
Sharad Chandra Sundararajan, Noohul Basheer Zain Ali, and Nikmohd Asrol AliasTo obtain a map of a restricted static environment by training an artificial neural network which translates the sonar readings to "probable" occupancy values and to use this map to navigate the robot.
Michael Boulet, Gregory Holden, and Gregory MillerGoal: To produce a complete configuration space (c-space) map of an initially unknown environment using SONAR range-finding sensors mounted on a mobile robot. Approach: The mobile robot, initially at rest in its starting position, builds a preliminary c-space map from data taken from its sonar sensors. The sonar data will be mapped to a rasterized representation of the configuration space using probability functions, similarly to the techniques discussed in reviewed research papers. The robot then analyzes the preliminary map to determine which areas of the c-space are either unknown or ill-defined (low certainties). It then attempts to find a location within the empty configuration space that will allow its sonar sensors to best resolve the unknown areas. The robot then plans a path and moves to the new sensor reading location using dead reckoning localization. (As an extension to the project we might try to incorporate sensor readings into the localization algorithm.) Once at the new location, the robot again takes data from its sonar sensors and combines it with the previous data to produce a more complete map of the configuration space. The robot continues to navigate around the configuration space taking sensor readings until a complete map is generated.
Chris Novak and Mary Kate WilliamsOur project would like to address the simplified case of a robot with a known map as well as sensors to detect one moving obstacle (and some rudimentary trajectory calculations based on path and speed). The problems of localization (vs. dead reckoning) will be addressed as well, most probably with a merging algorithm.
Rich Czyzewski, Dan Malone, Kelly Renny, and Matt SchumakerWe will create a software simulation of multiple robots navigating a fixed landscape with the goal of hitting all nodes. The simulation will be done using Java, creating objects for the robots, the landscape, and the graphical representation of the robots' motions. The landscape will be a 2-dimensional x-y coordinate system with polygonal obstacles. The robot object will implement a triangular motion platform of fixed size and speed. Our goal is to find the most efficient algorithm to control multiple robots in covering a landscape, so that the robots have sufficient room to maneuver around obstructions and each other.
Jason Shtrax and William von AchenWe would like to do a software project involving path planning for mobile robots. Specifically, we would like to develop algorithms to plan search patterns for mobile robots. The algorithm would obviously need to be applicable to areas of arbitrary shape and size. We would start by developing planning software which would plan a search pattern for one robot, then try to expand this to plan paths for multiple robots working in the same search area. Obviously, efficiency will be a key concern, since inefficient search patterns used in a search and rescue application could lead to failure (i.e., death of the person to be rescued!). Initially, the algorithm(s) could work on search areas with known geography; later improvements could possibly be made to have them work dynamically in areas with unknown geography. (Obviously, the boundaries of the search area would have to be specified; obstacles within the search area would be unknown.)
Leander Hasty, Michael Barrell, and Paolo CavalliTo implement a field-smoothing algorithm (primarily in C++) that can be applied to a potential field that would remove any local minima, which a robot might eventually use to navigate any imaginable navigable configuration space within software; and to test the reliability and robustness, as well as speed, of the particular implementation of the algorithm.