Thrust 3 Affiliated Project:

1
Thrust 3 Affiliated Project Optimal State
Estimators for a Flexible Robotic Arm Initial
Simulation Studies
Project Goals
Support of Strategic Plan

• Study which state estimator is the most useful
  and robust for application to flexible robotic
  arms.
• Design a controller to track the output of the
  state estimator and minimize tracking error.

Benefits of this work extend to long-reach
manipulators with hydraulic actuators. Examples
of existing hydraulic-powered flexible
manipulators in the IMDL include RALF and SAMII
mounted on a cantilever beam.
Results to date
The Problem
Various lumped parameter models with non-minimum
phase behavior have been computer simulated.
Limitations of over-simplified models were
discovered. Efforts have been focused on the
design and computer simulation of a stochastic
estimator, the Kalman filter. States including
tip position have been estimated based on
corrupted measurements from sensors. Tracking
performance of the Kalman filter has been
investigated. The metric used to judge
performance is mean square tracking error. The
closed-loop stability and transient performance
of the systems with the Kalman filter have been
studied. Surprisingly, the closed-loop system
remains stable despite very large parameter
uncertainty. This unexpected result is to be
further investigated.
Robotic arms are subject to bending, torsion and
axial compression. In certain applications, to
ensure accuracy and repeatability of the useful
end-point of the robotic arm, the flexible nature
of the arm must be taken into account during
design. Immediate benefits of this research
include
advisor
Tip Position
Wayne Book, GT
Motion of a Single Flexible Link
Time (sec)

• Improved control of long-reach space and
  lower-cost industrial manipulators.

• Improved accuracy and precision of general
  robotic manipulators with non-collocated
  actuators and sensors.

Tip Position Estimate using Kalman Filter
The Approach
Produce a useful model for a single flexible
link. Research and identify suitable state
estimators. Simulate estimator and tune estimator
parameters. Select an appropriate feedback
control scheme to supplement the estimator.
Benchmark the estimator performance by examining
robustness of closed-loop system to measurement
and parameter uncertainty.
student
Mohsin Waqar, GT
Tip Position
Lumped Parameter Model with Non-Minimum Phase
Behavior
Time (sec)
Closed Loop Simulation with Kalman Filter
Next steps

• More accurate modeling of non-minimum phase
  behavior.
• Identify other state estimators for study.
• Shift focus from simulation to experimental
  test-bed.

Explain the pictures if not obvious
Block Diagram for Experimental Test-Bed