Transcript IFAC SICICA2003

5th IFAC International Symposium on
Intelligent Components and Instruments for
Control Applications, Aveiro, 9-11 July 2003,
Portugal
Instrumentation and Localisation in Quasi-Structured Environments for AGV Positioning
Abílio Azenha and Adriano Carvalho
Institute of Systems and Robotics, Faculty of Engineering, University of Porto,
Rua Dr. Roberto Frias s/n, 4200 - 465 Porto, Portugal
E-mail: [email protected], [email protected]
Abstract:
This communication addresses the automated guided vehicles
(AGVs) positioning issue. A summarised state-of-the-art
section is included and the dead-reckoning algorithm is
analysed with a support on a triangulation scheme. This scheme
is based on ultrasonic sensors or electromagnetic wave
transmitter/receiver antennas sets. Simulation results for AGV
movement to validate the control system efficiency and
implementation issues of work in progress are presented.
Overview
Section 2 sketches a state-of-the-art overview summary.
Section 3 derives from WMRs technological knowledge to a model
for AGV positioning in quasi-structured indoors environments and
deals with the dead-reckoning algorithm and triangulation method.
Section 4 outlines the controller design and simulation results.
Section 5 introduces the current implementation issues.
Section 6 draws the main conclusions and future research work
issues.
STATE-OF-THE-ART
The development of a positioning system is based on the system
measurement requirements. Typically, different classes of
requirements according to either the vehicle and/or its movement can
be found.
global navigation, with ability to determine the object position in absolute or
map-referenced terms, and to move to a desired destination point;
local navigation, with ability to determine the object position relative to objects
(stationary or moving) in the environment, and to interact with them correctly;
personal navigation, which involves being aware of the positioning of the various
parts that make up oneself, in relation to each other and in handling objects.
STATE-OF-THE-ART
Automatic
warehouse
example with
AGVs
STATE-OF-THE-ART
•Dead-reckoning
•Odometry
•Global Positioning Systems (GPS)
•Inertial Navigation Systems (INS)
•‘Pseudo-satellites’ (pseudolites)
•Building Positioning System (BPS)
•Ultrasonic (or sonar) and laser (or lidar) sensor triangulation
•RF based triangulation algorithms
DEAD-RECKONING ALGORITHM AND
TRIANGULATION METHOD
Figure 1. Adopted WMR model.
DEAD-RECKONING ALGORITHM AND
TRIANGULATION METHOD
é
ù
é x1 ù ê x10 + R cos(f0 )(Dq1 + Dq 2 ) / 2ú
ê x ú = ê x + Rsin (f )(Dq + Dq ) / 2 ú
0
1
2
ê 2 ú ê 20
ú
R
êë f úû êf + (Dq - Dq )
ú
0
1
2
2b
ëê
ûú
where (x1, x2, f) is the current
WMR pose and (x10, x20, f0) is
the previous time step WMR
pose.
The triangulation method updates the AGV position from time to
time and an internal AGV compass updates its orientation. The
triangulation sensors analysed are ultrasonic and laser/RF beams,
based on distances and angles calculation triangulation algorithm
DESIGN AND SIMULATION
P2
(x1r, x2r)
i
u
ft
KI
Kt


WMR
q
1
s
Figure 3. AGV control scheme.
f
P1
(x1, x2)
Figure 2. AGV orientations model.
The control system attempts to
align f with ft. In this study the
control scheme is implemented by
the AGV micro-controller in a way
as depicted in Figure 3.
DESIGN AND SIMULATION
1 .1
x 2 r (m )
t = 2 s
t = 4 s
x 2 (m )
1 .2
1 .1 5
1 .0 8
1 .1
1 .0 6
1 .0 5
1 .0 4
a)
1
1 .0 2
0 .9 5
t = 0 s
t = 6 s
1
1 .1
1 .2
1 .4 x 1 (m ) 1 .5
1 .3
1
1
1 .1
1 .2
x 1 r (m ) 1 .3
0 .0 5
0
Figure 4. The AGV reference
trajectory.
x 1 e (m )
- 0 .0 5
b)
- 0 .1
x 2 e (m )
- 0 .1 5
Numerical Values:
R = 0.05 m, b= 0.2 m, l = 0.4 m, tw = 0.01 m,
mc = 10 Kg, mw = 0.45 Kg, Fvi = 0.5 Nms,
Kti = 31.1 mNm/A, KIi = 0.03 A/V and i = 66,
i = 1, 2
0
2
4
t im e (s )
6
Figure 5. a) AGV trajectory
response; b) AGV position
error signals.
IMPLEMENTATION ISSUES
Work in progress:
•localisation based on the 2.4 GHz band (ISM)
•chosen core micro-controller is the Atmel AT90S8535
•AVR-GCC freeware C compiler was adopted to develop the
program
•optical encoders are Hewlett-Packard HEDS-5540-A06 with a 500
points per revolution resolution
•PWM dc motors control bridges adopted are two Allegro
Microsystems integrated circuits A3952SB (one for each dc motor)
IMPLEMENTATION ISSUES
Work in progress:
•2.4 GHz transmitter based on a VCO (MAX2750)
•VCO output power is about -3 dBm and if rising that value is
wanted a PA such as MAX2240 should be adopted
•RSSI circuit based on AD8361 rms (root-mean-squared) converter
•AD8361 response is nearly linear to the input power and the output
is a quasi (slow varying) dc voltage. The absolute maximum input
power is 10 dBm at a matched antenna impedance of 50 W
CONCLUSIONS AND FUTURE WORK
•At present, a small WMR for position measurement and control
purposes is being built. Its electromechanical structure is already
implemented
•The trajectory is a priori known so the path planning is made off-line,
because the AGV will be placed in quasi-structured and flexible layout
indoors environments
•Triangulation algorithms with electronic beacons scattered
strategically around the quasi-structured indoors workspace are
planned to be used. It is expected that a scanning triangulation
frequency less than 1 Hz is sufficient for each moving AGV in the
environment, due to its low speed