V.

MACHINE EQUIVALENCE 2 TO BIOLOGICAL SYSTEMS

D. Tesar, The University of Texas at Austin, March 24, 2005

Objective: The ultimate goal of machine systems is to benefit from and integrate the intelligence structure in
biological systems to govern the motor capacity represented by machines (their control, dexterity,
reconfiguration, learning, self-healing, and refreshment) in order to perform complex human functions
(shooting a basketball 30 ft. from one’s fingertips within 100 to 300 milli-sec.) or to augment those
capabilities (precision, endurance, load capacity, etc.).
Background: Much has been written about the concept of artificial intelligence. Our present day computers
may still only be able to solve a simple set of logical decisions, but they are able to resolve truly complex
alternatives if given the correct norms, criteria, and processing algorithms. By contrast, the biological
equivalence in motor capacity (moving from complete softness in a delicate touch, to high forces in a
chopping action, to exceptional rigidity by using antagonism, to surgical precision using a specialized tool,
etc.) for machine systems is in its infancy. This lack of progress is primarily due to our inattention to the
essential ingredient between the computer and the physical task—i.e., the intelligent actuator. The actuator is
the exact equivalent as a driver of machines to the electronic chip as a driver of computers. The level of
standardization, depth of technology, investment strategy, etc. in electronics has accelerated during the past
30 to 40 years where truly unprecedented (also quite unbelievable) progress has occurred. The driving force
behind this progress was the continual benefit for measurable predicted (and self-fulfilling) progress on a
periodic basis and also the forecast as to how much further progress could occur with continued scientific and
commercial development.
The question is, what is the equivalent potential for intelligent Electro-Mechanical Actuators (EMA)? The
simplest perspective is to put EMA’s in a timeline similar to that which we have seen for electronic switching;
i.e., where electrical valves (analog tubes) were in 1950 is where EMA’s are in their potential development
today. The analog tube valve was standardized (but the exceptional population did not lead to rapid
technology progress, cost reductions, endurance, etc.), it had some standard plug-end connectors, it used
standard voltage levels, etc. but almost nothing more. This is where EMA’s are today. There are literally
thousands of devices on the market, each produced for its niche market. Their level of standardization is
minimal (interfaces haven’t changed for 50 years), integration (motors, brakes, buses, gear trains, sensors,
electronic controllers, etc.) almost doesn’t exist, their control is based on not failing (the only question is
stability), and a true sense of intelligence in a full architecture is non-existent in the trade. Yet, the worldwide
market for actuation devices is said to be $75 billion and growing at 50% every three years. The need is
there. The question is, what is the technical future for EMA’s?
Proposed Long-Term development: To achieve (more correctly to exceed) the intelligent motor capacity of
biological systems will require the same technical and investment strategy we have witnessed for computer
electronics over the past forty years. Every function of the human, or the support and augmentation of
humans, requires this improved motor capacity as represented by an intelligent EMA* Intelligence can only
occur in a set of actuators which have enough internal resources** to be managed by that intelligence. The
biological system can be reconfigured on demand (softness, rigidity, force, speed, etc.) to perform an
exceptional range of tasks. Today, our mechanical systems (say, industrial robots) are far removed from this
remarkably important goal. We must establish a finite number of actuator classes, make them intelligent,***
standardize their size and interfaces, manage their assembly on demand in completely reconfigurable systems,
operate them with one universal system software, permit human intervention, and continue to work the
performance/cost ratio. Doing so is not a dream (i.e., science fiction). It is a commercial opportunity
dependent on leadership and a commitment of adequate development resources.

1
Intelligence Dominates This Equivalence (See next page)
*
Attached, find a one-page description of our present concept of EMA intelligence.
**
The attached presentation chart describes the potential for diverse functional regimes in a full EMA architecture.
***
Attached is a presentation chart of the functional complexity necessary to make an actuator intelligent.

10
 ACTUATOR INTELLIGENCE
D. Tesar, The University of Texas at Austin
July 9, 2004

Purpose: The goal is to establish a fully responsive actuator whose intelligence manages a sufficiently
broad set of choices (performance, duality, layered control, force/motion, etc.) using carefully
documented criteria (for prime mover, bearings, gear trains, power supply, and electronic controller)
which when combined by fusion mathematics enables deployment to the widest range of systems
(aircraft, ships, battlefield, space, manufacturing, surgery, etc.).
Background: We have established a full architecture of 10 classes of rotary and linear actuators which
embody all the possible physical choices now considered necessary for an extremely broad set of
applications. We have energetically pursued the design of these actuators, and are formulating a science
of design for that purpose (parameter definition, design, criteria, configuration management, scaling
procedures, parameter reduction by synthesis, etc.). We have established strong position statements for
deployment of these actuators in virtually every application of mechanical systems. Because these
systems are nonlinear, the deployed actuators are highly coupled, and the actuators themselves are highly
nonlinear (perhaps 30 criteria are necessary to describe their operation), it becomes necessary to develop a
specific scientific approach to manage these actuator resources by means of criteria fusion, ranked and
normalized criteria (prioritized based on their physical meaning and relevance), with priority setting done
primarily by human judgment.
Technical Development: Now that we have a fully established actuator architecture, a well defined set
of operational criteria, and an emerging decision making process, the underlying science can flourish and
we can work towards the following:
1. Maximum Performance The most basic need for intelligence is to combine available resources
Envelope within an actuator which provides the best performance envelope to meet
(See Jae Yoo Report) the present needs of a given application (force, accuracy, speed, response,
etc.) on demand.
2. Condition-Based The broadest possible early need for actuator intelligence will be the
Maintenance condition monitoring of the actuator to advise the system if the performance
(See P. Hvass Report) envelope has diminished and whether maintenance is required.
3. Fault Tolerance Where life is a stake (aircraft) or where high economic losses might occur
(See Aircraft Safety (nuclear reactors), then fault avoidance becomes necessary. Here, we have
Proposal) equal (dual) subsystems (performance maps) which must be balanced in real
time to ensure continued operation under a fault (full or partial) on one side.
4. Layered Control Here, we mix the physical scales of the system (say 100%, 10%, 1%, etc.).
(See CHAMP proposal) Unique criteria exist at each scale, mixing criteria between scales must be
developed, and performance objectives (norms) must be set at each scale.

5. Force/Motion Control Here, we mix physically distinct phenomena to enable a whole new class of
(See D. Rabindran work) output functions to be met where a basic motion must be achieved without
being impacted by a superimposed force disturbance.

The criteria for this intelligence are built on the physical nature of electro-magnetic prime movers, rolling
element bearings, tooth mesh reducers, and electronic power supplies. Using other basic components
(piezoelectric drivers; jeweled, air or magnetic bearings; or screw transmissions or fluid reducers) would
add a new range of criteria to our concept of actuator intelligence.

11
FUNCTIONAL REGIMES FOR ELECTRO-MECHANICAL ACTUATORS

1. CREATES • High Gear Ratio, R > 100

FORCE – Low/Undisturbed Output Velocity
(FA) – Reduces Overall Actuator Weight ≈50x
– High Load Capacity/High Output Stiffness
– Resists Gravity/Holds Position Well
– High RPM Responsive Motor
– Motor Sees Virtually No System Inertia
2. CREATES • Low Gear Ratio R<20
VELOCITY – Manages Downstream Force Levels
(VA) – Provides High Output Acceleration
– Modest Output Stiffness/
Hard Surface Contact Feasible
– Output Inertia Affects Operation
3. FORCE AND • Separate Force and Velocity Outputs
VELOCITY – Doubles Number of Input Commands
(FVA) – Expands Output Task Generality
– Coupling May Occur At System Level
4. FORCE • Dual Equal Force Level Actuators
SUMMING – Provides Fault Tolerance
(FFA) – Forces Add/Reduces Weight 2x
– Requires Output Release Clutches
5. VELOCITY • Dual Equal Velocity Level Actuators
SUMMING – Provides Fault Tolerance
(VVA) – Velocities Add/Doubles Output Power
– No Clutches Required
6. LAYERED • Two Scales of Force in Series
FORCE – Each Has High Output Stiffness
(FfA) – Large Scale Uses High Ratio R
– Small Scale Uses > 10R
– Small Scale May Use Piezo-electrics
– Rejects Output Disturbances
7. LAYERED • Two Scales of Velocity In Series
VELOCITY – Mixes Two Levels of Acceleration
(VvA) – Manages Output Force Levels
– Creates Desired Force Output ”Jitters”

12
 INTELLIGENT ACTUATOR
• Array of Available Resources
• Distribution of Interfaces
• Criteria Based Performance Management

13