An Information Fusion
Approach to Missile Guidance
• Name
Yaakov Oshman
• Affiliation
Technion - Israel Institute of Technology
• Abstract
We address the problem of guiding a
missile towards an evading agile target in the presence
of noisy measurements. State-of-the-art interception missiles
derive their superior performance mainly from highly advanced
aerodynamic configuration design, agile seekers and highly
efficient fragmentation warheads, but use only relatively
basic information on the state of the target. Recent advances
in onboard computing and imaging technologies render the
expansion of this information base feasible, thus enabling
the use of advanced guidance laws.
We focus on an integrated, information fusion approach to
missile guidance, which facilitates synergy between the
target estimation and guidance functions of the missile
autopilot. As a first example, we examine the advantages
of fusing classical line-of-sight observations with target
orientation (attitude) observations, obtained by implementing
computer vision techniques on visual measurements acquired
by high-resolution imaging sensors. We show that, by properly
exploiting the various sources of information, and using
the target's bank-to-turn mathematical model, the missile's
closed-loop interception performance can be vastly improved
even when using an off-the-shelf differential game-based
guidance law, which is derived based on an unrealistic perfect
information assumption.
Next we make a bolder step in the direction of estimation/guidance
fusion. Existing missile guidance law design methods traditionally
assume the validity of the celebrated separation theorem.
This assumption permits a separate design of the optimal
estimation and guidance (control) functions, while retaining
the optimality of the entire guidance system. Alas, although
this approach is feasible and leads to many practical and
working designs, the obvious problem with it is that the
separation theorem has never been proven valid in realistic
(nonlinear, non-Gaussian) interception scenarios. In such
cases, only a generalized separation theorem may be applied,
implying a separately designed estimator, but a guidance
law that takes into account the conditional probability
density function generated by this estimator. Although the
generalized separation theorem has been known since Witsenhausen's
1971 groundbreaking article on separation of estimation
and control for discrete time systems, little has been done,
so far, in this direction.
In this talk we present a new approach to fusion of estimation
and guidance, under the guidelines of the generalized separation
principle. To this end, the conventional notion of reachability
sets is extended, introducing the notions of pursuer and
evader miss-sets. Adopting a geometry-based approach the
necessary conditions for guaranteeing a hit are derived
in the general case of nonlinear dynamic models without
constraining the analysis to the standard Gaussian noise
assumptions. A nonlinear, non-Gaussian numerical study is
presented, that demonstrates the performance of the proposed
methodology in a 3-D realistic engagement scenario with
partial information. Particle filtering is used to approximate
the entire state conditional probability density function
using the exact nonlinear dynamic models without resorting
to the standard Gaussian noise assumptions.
• Plenary Talk II
Receding Horizon Schemes
for Controls, Estimations, and Optimizations
• Name
Wook Hyun Kwon
• Affiliation
Seoul National University - President
of IFAC
• Abstract
Major achievements of receding horizon
concepts applied to control, estimation, and optimization
problems for state space models, hybrid models, and Markov
decision processes will be surveyed. Mathematical description
of these problems will be presented in a general framework
of minimization and mini-maximization optimizations.
For control problems, we first introduce the receding horizon
controls (RHC) for state space models with their advantages
and properties. The receding horizon controls are obtained
from LQ and criteria as in infinite horizon controls. As
an important property of the RHC, the cost monotonicity
condition is discussed for nonlinear systems. For the partial
information on the state, LQ and finite impulse response
(FIR) filters and output feedback receding horizon controls
(OFRHC) are introduced for linear systems. Specially, we
introduce the separation property that the overall systems
are composed of the state feedback RHCs and the FIR filters.
Popular generalized predictive controls (GPC) for I/O models
are also introduced as output feedback controls and compared
with the LQ OFRHC for state space models. It is shown that
RHCs for unconstrained systems can be easily extended to
constrained systems due to the receding horizon scheme.
Useful invariant sets for handling the input and state constraints
will be covered and how to represent a form of linear matrix
inequalities (LMIs) will be dealt for numerical programming.
For state estimation problems, receding horizon filters
for piecewise-affine hybrid systems are introduced. The
stability condition with respect to the horizon size and
the weighting on the estimated state at the initial time
of the receding horizon will be discussed with a connection
to cost monotonicity. Algorithms for the computation will
be discussed.
For optimization problems, a receding horizon approach is
introduced for Markov decision process. Best policies for
one person average rewards and two person zero sum games
will be reviewed and their properties are shortly discussed.
Above two criteria for Markov decision process will be compared
with LQ and ones for state space models in view of minimization
and mini-maximization.
• Plenary Talk III
New Actuators for Advanced
Mechatronics
• Name
Toshiro Higuchi
• Affiliation
The University of Tokyo
• Abstract
Applications of actuators have increased
in various fields. In industry, precise and high speed positioning
is one of the most important technologies. In peripheral
machines for computers like disc memories, small and thin
actuators are necessary to satisfy the demand of reduction
of thickness and weight of the products. In robotics development
of powerful and light-weight actuators is a key issue to
improve dynamic performance and to reduce the total weight
of robots. In some cases conventional actuators seem to
be difficult to satisfy these new and advanced demands.
Therefore, seeking for new actuators has been activated
recently. Especially, in the field of MEMS, where the dominant
physical rules are different from those of the usual size,
new small actuators utilizing electrostatic force, heat
deformation, piezoelectric materials, shape memory, and
so on have been proposed and developed.
In my laboratory, a number of unique actuators and drive
technologies were invented and realized. Among them, first,
the following two actuators using piezoelectric materials,
impact drive mechanism and surface acoustic wave motor,
are introduced with applications. Impact drive method can
move an object with a step of several nm to micrometers
for a long distance by using rapid deformation of a piezoelectric
element. The surface acoustic wave motor is a very thin
ultrasonic motor with promising properties like large thrust,
high velocity, and quick response. Then, as applications
of electrostatic force, powerful electrostatic motor and
new technologies of electrostatic transportation of particles,
powder and droplets, sheets and films, and thin plates,
are presented respectively with their principles and devices.
And combining the technologies of electrostatic suspension
without mechanical contact and electrostatic drives, super
clean transportation devices for 300 mm and 400mm silicon
wafers and thin plates of glass for flat displays are introduced.
Organized by ICROS(Institute of Control, Robotics and Systems)
Bucheon Techno Park 401-1506, 193, Yakdae-Dong, Wonmi-Gu, Bucheon-City, Gyeonggi-Do, 420-734, Korea
Phone : +82-32-234-5801/ Fax : +82-32-234-5807/ E-mail : conference@icros.org
