CATALOG DATA:
A study of tools of estimation and stochastic modeling and their use in the application of artificial vision to the guidance and control of multi-degree-of-freedom mechanisms. The Kalman filter and extended Kalman filter are developed; state and observation equations, based, respectively, on robot mechanisms and discrete visual data. Issues of image analysis, time delay, and the modeling of random-disturbance covariances as well as kinematic holonomy.TEXTBOOK:
noneGOALS:
The goal of the course is to convey both the promise and the difficulties of controlling mechanisms using computer vision. Emphasis is placed upon Camera Space Manipulation as a robust and useful approach for 3D control of both holonomic and nonholonomic robots. Students are expected to emerge from the course with first-hand knowledge of the many programming, mechanism, image analysis and error-analysis aspects of what is required to design and build a working robotic hand/eye coordination system.PREREQUISITES:
AME 50551 (AME 469) Intro to Robotics. Experience programming preferably in C++ or closely related language.Topics:
- Holonomic and nonholonomic robot kinematics
- Probability density functions
- Extended Kalman filter
- Optics: perspective camera model and real multiple-element lenses
- Camera-space manipulation
- Image acquisition/digitization
- Robot control topics, brief historical perspective: Teach-repeat,calibration, visual serving
ABET category content as estimated by faculty member who prepared the course description:
Engineering Science: 1.5 credits or 50%
Engineering Design: 1.5 credits or 50%
Prepared by: Professor Steven Skaar
Last Update: May 14, 2004
Direct comments, questions, and corrections to amedept@nd.edu