Title: Motion Planning of Robotic Systems for Applications in Nuclear Facilities Clean Up
1Motion Planning of Robotic Systems for
Applications in Nuclear Facilities Clean Up
- By
- Peter S. March, Chetan Kapoor, and Delbert Tesar
- Robotics Research Group
- The University of Texas at Austin
2Presentation Outline
- Introduction
- Motion Planning (MP) Software
- Reasons for MP Software
- Requirements for MP Software
- Development of MP Software
- Applications of Software
- DD Demo
- DD Plasma Torch Cutting
- Glovebox Automation
- Conclusion
3Robotics Research Group (RRG)
To produce modular systems which exhibit
advanced performance at reduced costs whose
architecture matches that of todays computers,
allowing rapid repairs and a reduced threat of
obsolescence
- Two Major Areas of Research
- Intelligent Actuators for Robotics
- System Level Decision Making and Software
- Accomplishments
- 50 M.Sc. and 20 Ph.D. Graduates since 1985
- Currently, 12 M.Sc. and 7 Ph.D. Students
- More than 17M in funding since 1985
- Sponsors have included NASA, DOE, ATP, DARPA
4Introduction
- University Research Program in Robotics (URPR)
- DOE Sponsored Program Conducting Research on
Robotic Applications for Nuclear Facilities Clean
Up - Participating Universities
- University of Florida
- University of Michigan
- University of New Mexico
- University of Tennessee
- University of Texas
5Introduction
- Why Use Robotics?
- Reduce onsite involvement of humans in hazardous
environments - Reduce Overall Costs
- Problems with Robotic Use
- Difficult to implement
- Difficult to maintain
- Requires highly skilled operators
6DD Tele-Robotic Task
- Concept
- Move away from pure tele-operation and pure
automation to perform more complex operations - Combine tele-operation, sensing, automation,
motion planning, decision making, etc
- Task Level Planner
- Decision-making
- Trajectory Execution
- Data Input
- Tele-operation
- Vision Systems
- Sensing
- Motion Planner
- Trajectory Generation
7Motion Planning Software
- What is Motion Planning?
- Motion Planning is the generation of smooth
motion trajectories for robotic systems. - Types of Motion Planning
- Joint-Space Planning
- End-Effector (Cartesian)
- Space Planning
(xf,yf,?f)
(xi,yi,?i)
8Motion Planning Software
- Trapezoidal Motion Specification
- Define derivative of motion curve to have
trapezoidal shape - Derivative that trapezoid is defined in is known
as the order of the system - Positions determined through integration
2nd Order Motion System
9Motion Planning Software
- Software Requirements
- Easy to use and reprogram
- Applicable to a variety of robotic systems
- Able to produce trajectories in both Joint Space
and End-Effector (Cartesian) Space - Able to produce trajectories with respect to
defined constraints (max velocity, max
acceleration, etc.) - Able to be easily expanded to more complex
operations
10Motion Planning Software
- Operational Software Components for Advanced
Robotics (OSCAR) - Concept
- Generalized to N-DOF Systems
- Criteria Based Decision Making Core
- Open Interfaces via Emerging Technologies
- Real-Time, Simulation, and Design
- PC Based
- Development
- 3 developers for past 5 years
- Online reference manual with tutorials and sample
code
http//www.robotics.utexas.edu/rrg/downloads/softw
are/oscar/
11Motion Planning Software
OSCAR Implementation Overview
Operator Interface Layer
Machine Interface Layer
Generalized Operational Software Layer
- Forward Kinematics
- Inverse Kinematics
- Performance Analysis
- Obstacle Avoidance
- Dynamic Modeling
- Deflection Modeling
- Motion Planning
- Real-time Event Handling
- Hardware Interfacing
- Simulation Interfacing
- Servo Communication
- Network protocols
- Robot Programming
- Variety of controllers.
- Variety of operator backgrounds.
- Compatible with GUI Programming Objects
- Network protocols
12Motion Planning Software
- Class Hierarchy for MP Software
- MotionPlan
- Contains routines to
- generate trapezoidal
- motion profiles
- OffLineMotionPlanner
- Generates and stores
- trajectories (useful for
- automation)
- OnLineMotionPlanner
- Generates real-time
- trajectories (useful for
- tele-operation)
- MotionParameter
- Stores information on time intervals,initial and
final configurations, and constraints (max
acceleration, max velocity, etc)
MotionPlan
OffLineMotionPlanner
OnLineMotionPlanner
MotionParameter
13DD Demo
- DD Demo
- Used as validation of software
- Robotics Research Corporation K/B 2017 17 DOF
dual-arm robot - Performs a demonstrative DD task (material
reduction) - Uses Joint Space and End-Effector Space planning
- Executed with linear interpolation, and 1st and
2nd order motion planning
14DD Demo
- 5 Motion Curve Criteria
- ymax - maximum velocity of the motion, provides
a good indicator of the inertial energy - yrms - root mean square of the acceleration
curve, high values are an indication of high
inertial forces - ?y - peak-to-peak value of the acceleration
curve, indicates the range of inertial loading. - ?? - peak-to-peak value of the torque curve,
indicates the range of torque demands. - ?max - derivative of the torque curve.
15DD Demo Results
- Easily integrated into existing application
- Visual improvement over linear interpolation
routines - Improvement in all
- criteria except
- max velocity
- Improvement
- most dramatic
- with joint space
- planning
- 1st Order planning
- showed more
- improvement than
- 2nd Order planning
-
-
Joint Space Planning
End-Effector Space Planning
16DD Demo
17Application DD Plasma Torch Cutting
- Plasma Torch Cutting Simulation
- Schilling Titan 2 with 1-DOF End-Effector
- Tele-Robotic Operation
- Tele-operated to
- determine set points
- Automated task
- performs cut
- Decision-Making
- Determination of Tool configuration
- Optimization of rotation about z-axis of tool
18Application DD Plasma Torch Cutting
19Application Glovebox Operation
- Modular Small Automation Systems
- 1 DOF to 5 DOF Systems
- Manually Reconfigurable Through Glove Port
- Motion Planning for Efficient Paths and
Automation - Physical Demonstration
- DOE URPR and LANL Supported Research
20Conclusion
- Software Results
- Easily integrated into existing application (DD
Demo) - Extensible to more complex applications (Plasma
Torch Demo) - Resulted in visibly smoother motion and improved
system performance over linear interpolation - Future Work
- Addition of Online motion planning
- Applications to more complex tele-robotic tasks
- Improved end-effector space planning