Transcript Slide 1

Antonio Yordan-Nones
Heterogeneous Modeling & Design of a
Robot Arm Control System for Ptolemy II
University of Puerto Rico,
Mayagüez
Mentors:
Prof. Edward A. Lee
Xiaojun liu
http://chess.eecs.berkeley.edu
Motivation
Robots and embedded control systems are
traditionally programmed to perform automated
tasks, yet accomplishing dynamically changing
tasks will typically require the use of specialized
software systems. In order to design these
sophisticated control systems the use of
computational
modeling
and
simulation
techniques is necessary to ensure that final
implementations of the system perform
optimally. Since even the simplest robotic
control systems are intrinsically a mixture of
heterogeneous sub-systems, each focusing on a
particular task, we need a means of modeling
the different aspects of the system accurately
and intuitively.
Designing a Robot Arm Actor
Input:
•3D Cartesian Coordinates
Output:
•Incremental Servo Positions
•Notify reaching outside of the
robot’s workspace
Design:
•Calculate the inverse
kinematics
•Convert the degrees of rotation
to absolute servo positions
•Verify servo bounds to stay
inside of the robot’s workspace
•Step through the sequence of
movements / positions of
the servo motors
•Output a stream of bytes to the
external serial servo controller
The focus of study in the Ptolemy project is on
modeling, simulating, and designing embedded
systems that require heterogeneous and hybrid
modeling, component concurrency, and realtime
responsiveness.
Ptolemy
extends
component-based
design
principles
by
accurately managing the time continuum and
maintaining concurrency among actors, features
that are especially useful when designing
control systems software.
August 7, 2003
Modeling embedded control systems for
robotics in Ptolemy is very flexible since there
are a variety of models of computation to choose
from. For example, an electro-mechanical system
could be represented under continuous
dynamics, and a digital communications system
may be more appropriately modeled under the
discrete-event domain. Systems that employ
multiple modes of operation with overlapping
computational domains are hybrid models.
Lynx-5 Robot Arm
CASE STUDY:
Lynx-5 meets X-10 Remote Control
PtolemyII: An Overview
Modeling Control
Systems in Ptolemy II
Features:
•A novel embedded system:
•concurrent components
•discrete-events (DE)
•X-10 wireless remote control
sends commands through the
X-10 network
•Specify the cardinal orientation
to move the arm
( up - down - left - right )
•Implement communication
between 2 different models &
2 different computers by
sending datagrams
To apply modeling techniques in Ptolemy and
demonstrate embedded software control of a
robotic manipulator we chose the Lynx-5 Robot
Arm. The Lynx-5 is an articulated manipulator
(RRR) with 2 planar links in an elbow-like
configuration and 4 degrees-of-freedom - each of
the 4 joints are controlled by dedicated servo
motors.