|
| Research Projects |
Teleoperation and Automation |
|
|
| Overview |
| Our research in teleoperation and automation consists of two
threads: Human Machine Interface and Manual Controllers. |
|
| Human Machine Interface |
In any robotic system, the interface between human and machine
plays an enormous role in the success of the system. In industry, this
interaction is often very basic because of the high level of automation that
can be preprogrammed into robots in structured environments. However, in
unstructured applications where humans continuously interact with robots the
interface is much more complicated because the environment is unknown and
dynamic.
Typically, this problem is approached via teleoperation with a joystick or
other suitable input device. Rarely, however, is this enough to fully control a
system. A graphical user interface (GUI) of some type is almost always
developed to accompany the input device. Also, new techniques of human to robot
communication are constantly being developed including voice and gesture
recognition, GUIs, et cetera. Much of this development of an interface for a
given robotic system must then be repeated when a new system is introduced.
RRG proposes a robot independent
architecture for human-robot interaction that supports integration of various
input devices and resolution of resulting conflicts. The architecture will also
allow for these same interfaces to be ported to various robots with minimal
changes. Though the focus will lie heavily in the field of
robotics, many of the principles will be applicable to other domains where
considerable control is left to the operator.
|
|
| Manual Controllers |
With the prevalence of teleoperation in robotics application today,
manual controllers (MC) are an integral part of the robotic system. While some
systems eliminate MC input with enough automation or other interface types, the
overwhelming majority still use MCs. Even industrial robots have simple teach
pendants that allow incremental movements in joint or end-effector (EEF) space.
Manual controllers can be categorized into four basic types:
- Joint-Level
Joint-level manual controllers include any device that is intended to simply
command the joint position of the robot. The most common types of joint-level
MCs are lever-boxes or button-boxes.
- Kinematic Replica
Kinematic replica MCs are very common in robotic teleoperation tasks where the
robot is intended to mimic the operator’s movements exactly. They are kinematic
arms usually modeled after a robot arm for which they were intended to serve as
a master.
(TOS, Kraft)
- Universal
Universal manual controllers are kinematic devices designed to optimize
performance within a particular work envelope. They are useful for both delta
and Cartesian control. Most research at RRG focuses on universal manual
controllers.
(PerForce)
- Delta
Delta Controllers have a small range of motion and a fixed base that make them
suitable only for delta control. They also have a fixed zero position allowing
them to easily supply a delta value, though it must usually be scaled for the
application.
(SpaceBall)
Various Manual
Controller Images
|
|
| Controller Software |
| OSCAR Version 2.0 |
|
| Demonstrations |
| Demo One:
Prototype Demonstration with K/B 2017 Dual-Arm Robot |
| Demo Two:
Motion Planning of Robotic Systems for Applications in Nuclear Facilities Clean
Up |
| Demo
Three: Decision Making for Deactivation and Decommissioning Robotic
Applications |
| Demo
Four: Teleoperation and Automation Demo |
|
| Demo One: Prototype Demonstration with K/B 2017 Dual-Arm Robot
|
|
|
|
| Demo Two: Motion Planning of Robotic Systems for Applications in Nuclear Facilities Clean Up
|
| Robotic systems are often used in the clean up of nuclear
facilities due to the uncertain environments and high levels of radiation. The
wide variety of tasks involved in nuclear facilities clean up will require that
systems constantly be reprogrammed as tasks are added or changed. This creates
the need for a generalized motion planning software capable of being applied to
a wide variety of systems and tasks. |
|
| This paper describes the basics of motion planning and the
development of motion planning software. The motion planning software is then
applied to a demonstrative Deactivation and Decommissioning (D&D) task to
illustrate enhanced performance. |
| Full Report |
|
|
|
| Demo Three: Decision Making for Deactivation and Decommissioning Robotic Applications
|
|
Presentation |
|
Report |
| Video |
Description |
| Video 1 |
A simulation of the Titan II robot with the 1-DOF Plasma Torch
Tool performing cuts on an angle bracket. |
| Video 2 |
The real Titan II system simulating cuts on a piece of poster
board. |
| Video 3 |
The Compact Remote Console (CRC) being used to remotely operate
the system |
| Video 4 |
A telerobotic simulation of the Pit Viper system performing a
wall-cleaning task. |
| Video 5 |
A simulation of a method for intelligently placing the backhoe
for optimal performance in a region of the pit. |
|
|
| Demo Four: Teleoperation and Automation Demo
|
|
The objective of this work was to demonstrate a novel approach to human machine
interaction that seamlessly uses teleoperation and automation in a complex
environment. This work leverages our development in the area of operational
software (OSCAR), decision making, human-machine interface, and motion
planning. This demonstration uses a 17 degrees-of freedom dual arm robot that
is equipped with modern tool changers, crash protectors, force torque sensors
and electrical and pneumatic power at the tools.
|
|
| Four different end-effector tools are also provided. These are
electric grippers, electric rotary saw, electric drill, and a pneumatic spray
gun. The system can be used both in teleoperation and automation mode. In
teleoperation mode, the user has a choice of five different input devices.
These are computer keyboard, Spaceball and SpaceMouse, RSI manual controller
and Kraft force feedback controller. Automation is performed using a novel
graphical user interface with 3D graphics used for previewing and verifying
manipulator motion. Automation tasks that are demonstrated include automatic
grasping, sawing, drilling, spray painting, point-to-point motion, and
teaching. The controller for the dual arm system is developed using OSCAR and
supports a variety of decision making algorithms and obstacle avoidance.
|
|
| The integration of this controller with the input devices and
human machine interface is done using a novel protocol that is based on XML for
maximum reuse and distributed integration. This protocol is further based on a
well-defined and scalable XML schema that can be easily extended as controller
functionality is changed and/or additional input devices are added. |
|
Full
Report |
| Pictures |
| Videos |
| Publications |
|
|
|
|
