Automation of Autonomous Largescale Production

Systems
V.M. Chadeev N.I Aristova
Trapeznikov Institute of Control Sciences RAS Trapeznikov Institute of Control Sciences RAS
Moscow, Russia Moscow, Russia
chavama@ipu.ru avtprom@ipu.ru

Abstract—The paper covers the procedure how to choose the Let us calculate the scheme of automated autonomous
most suitable (based on the number of workers in the production using the given inputs; such a scheme will allow to
sustainability system) automation scheme in autonomous minimize the costs borne by the people.
production systems. An abstract of model of production is used,
such model being invariant with the types of technological II. PARAMETERS OF THE PRODUCTS AND ROBOTS OF THE
operations. Three levels of automation at an autonomous station AUTONOMOUS SYSTEM
are in focus: Fully manual assembly of products with no
automata. Product automated production by means of purchased
A. Parameters of the products
automata. Automated production of both products and automata Parameters of the products to be manufactured within the
used for the product manufacture. This paper also highlights the system are determined by the following matrices:
algorithms of determining the optimal sorting of the types of
robots by the varieties of technological operations.
n11 n 21 ! nk1 ! nS1
Keywords—self-reproduction of robots, automation, level of
n12 n22 ! nk 2 ! nS 2
automation, operation suport system, technological operation,
operation performance time, production theory invariant with the ! ! ! ! ! ! (1)
types of technological operations n = n ki =
n1i n2i ! nki ! n si
I. INTRODUCTION ! ! ! ! ! !
Nowadays any production is designed to a certain degree n1m n2 m ! nkm ! n Sm
and from the very beginning as automated. Such production
may be achieved using a rich assortment of automation tools
which vary in cost and functionality [1-4]. N = N k = N1 N1 ! N k ! Ns , (2)
In certain circumstances the automated production is quite
a critical issue. It may be the case of autonomous systems, such D = d k = d1 d2 ! dk ! ds , (3)
as polar and orbital stations. Station workers are to be able to
produce any damaged component, therefore the station must
where nki is the number of i-type TOs that is needed to
have its own production. The automated production in such
systems is aimed to minimize the dependence on the Earth and manufacture one k-type product, S is the number of types of
products, Nk is the number of k-type products to be
to import as rarely as possible.
manufactured, dk is the cost of purchased products needed to
An autonomous production system may be described if the manufacture one k-type product.
following information is available:
B. Parameters of the robots
- a list of all products manufactured within the system Parameters of the robots used at the station are determined
(including robots), by the following matrices:
- description of the production procedure for each type of
the products and the robots,
b11 b21 ! b j1 ! b f 1
- specifications of the robots as elements of production, b12 b22 ! b j 2 ! b f 2
- a list and specifications of all types of technological ! ! ! ! ! ! (4),
operations (TOs); such a list and specifications are needed for b = b ji =
production of all types of the products and robots, b1i b2i ! b ji ! b fi
- a total number of people in the autonomous system. ! ! ! ! ! !
b1m b2 m ! b jm ! b fm

978-1-7281-1730-0/19/$31.00 ©2019 IEEE

Authorized licensed use limited to: University of Liverpool. Downloaded on July 26,2020 at 23:44:22 UTC from IEEE Xplore. Restrictions apply.
 where bji is the time during which i-type TOs are performed
by a j-type robot, f is the total number of types of robots used a11 a 21 ! a j1 ! a f 1
in production. a12 a 22 ! a j 2 ! a f 2
! ! ! ! ! !
h11 h21 ! h j1 ! h f 1 A = a ji = (11),
a1i a 2i ! a ji ! a fi
h12 h22 ! h j2 ! h f 2
! ! ! ! ! !
! ! ! ! ! !
h = h ji = (5) a1m a 2 m ! a jm ! a fm
h1i h2i ! h ji ! h fi
! ! ! ! ! ! αij is the level of automation of i-type TOs performed by a
j-type robot. The level of automation must meet the following
h1m h2 m ! h jm ! h fm conditions:
where hji is the number of i-type TOs needed to
manufacture one j-type robot. 0 £ a ji £ 1 ; i = 1, 2,..., m ; j = 0,1,..., m . (12)

The service life of robots (T), the cost of robots (C) and the
If all i-type technological operations are performed by an
cost of the working time unit of robots l [5] are input by the
automaton then the level of automation αi = 1, and if by a
following matrices, where j-index stands for the type of robot:
person - then 0. The following equality should be applied to all
types of TOs:
T = T j = T1 T2 ! T f , (6)
f

C = C j = C1 C2 ! C f , (7)
åa
j =0
ji = 1 ; i = 1, 2,..., m . (13)

Product manufacture cost G(А) is a function of the

l = l j = l1 l2 ! l f . (8) automation matrix. The optimal automation matrix A contains
only 0 and 1 and minimizes the production cost. This means
C. Human capacities that the matrix components a are on the boundary of the field
of definition (12). And each line of Matrix A contains a non-
Production capacities of people are contained in the zero element as per (13).
following matrix:
III. PRODUCT COST CALCULATION
G = gi = g1 g2 ! gm , (9) Cost of product automated manufacture by a system of
robots with the use of the automation matrix (11) is calculated
where gi is the time needed for a person to perform an i- according to the formula [6]:
type TO. To shorten the formulae and for the sake of m m m
convenience in some places we will regard a person as a 0-type G k (a ) = å a 0i b0i l0 nki + å a1i b1i l1nki + ! + å a fi b fi l f nki
robot. Then instead of (9) we will get: i =1 i =1 i =1 (14)

b0 = b0i = b01 b02 ! b0 m . (10) where m is the number of types of TOs required to
manufacture a product, bij is the time needed for a j-type
automaton to perform i-type TOs, lj is the cost of the working
A person's working time T0(t) will be an aggregate of the
time unit of a j-type automaton, gi is the time that a person
person's working time while working at the station t (for a year,
needs to perform i-type TOs, f is the total number of the types
for instance).
of robots available for automation, nij is the number of identical
II. AUTOMATION MATRIX i-type TOs needed to manufacture a product Gk(a).
The planning of automation of autonomous production is In (14) the first amount stands for the human costs, the
aimed to distribute the operations between people and robots, second amount stands for the first-type robot costs and the last
so that people could save on costs as much as possible. This amount stands for the f-type robot costs. If we write it
requires the use of an automation matrix as a management tool: laconically, we will get a general formula of calculation of the
product cost [6-8]:
f m
Gk (a ) = å åa ji b ji l j nki . (15)
j =0 i =1

Authorized licensed use limited to: University of Liverpool. Downloaded on July 26,2020 at 23:44:22 UTC from IEEE Xplore. Restrictions apply.
 The cost of the manual manufacture of the product is the N0=G/T0, (19)
starting point for all calculations made to assess the automation
efficiency. In this case, (15) will look simpler: where T0 is the worker's capacity per year.

m
The next step is to define the production automation
Gk (0) = å gi nki , (16)
scheme. The automation of the product manufacture is aimed
to reduce the number of workers supporting the operation of
i =1
the station. This goal is achieved in several stages. The first
stage means searching for the optimal automation vector.
where nki is the number of i-type TOs needed to
manufacture one k-type product, gi is the time required for a B. The optimal automation vector
person to perform i-type TOs. The optimal automation vector must be chosen in order to
pinpoint the type of robot that will be the most efficient in
IV. DATA FOR THE CALCULATION OF AUTONOMOUS
performing the given TO.
OPERATION SUPPORT SYSTEMS
Initial data required for the calculation should contain The first selection criterion is the time needed to perform a
information on the products that will be used at the station, on TO. Among all types of robots only the one that is the fastest to
the technology of their production and on the available means perform the given operation will be selected.
that will make their production automated. These data can be The second selection criterion is the cost of performance of
found in lists. a TO.
1. List 1 with the length of S-lines for all types of products The method allowing to calculate the automation vector
needed for the operation of the station. based on the criterion of the cost of performance of a TO is as
2. List 2 with the length of m-lines for all types of TOs follows.
needed for the manufacture of each S-type product from list 1. First we calculate the cost of the working time unit for all
3. List 3 with the length of S-lines containing the sizes of types of robots:
batches of each type of the products needed for the operation of
the station during T years. l j = C j / T j , j=0, 1,...,f. (20)
4. List 4 with the length of f-lines for the available types of
robots that are able to perform all or some TOs from list 2. Then we build a cost matrix when all types of robots with
the elements perform all types of TOs
5. Table of production parameters of robots from list 4
(cost, service life, time required for performing TOs from list
2). X = xij = bi l j , i=1, 2,...m; j=0, 1,...,f. (21)
6. Characteristics of the workers who will manufacture the
products from list 1 (working time, time needed to perform From each line, related to a certain type of a TO, of matrix
TOs from list 2). A the minimal value of the cost of performance and the type of
robot corresponding to this column are taken according to the
V. METHOD OF CALCULATION OF AN AUTONOMOUS following algorithm.
OPERATION SUPPORT SYSTEM

A. The required number of the workforce zi = min < xi 0 xi1 ! xif > , i=1, 2, ...,m, (22)
The autonomous station will need the maximum number of
the workforce if all TOs needed for the manufacture of where zi is the minimal cost of performance of an i-type TO
operation support products are performed by the workers. Let by one of the (f+1) types of robots.
us calculate the number of workers that will support the
operation of the station. For this purpose, we will calculate first The optimal automation vector is formed
the cost of manual manufacture of each S-type product
Z ( x ) = Z ( z1 z 2 ! zi ! z f ) , (23)
Gk = å g i nki ; k=1, 2,...,S. (17)
i where zi is the minimal cost of performance of an i-type TO
by one of the (f+1) types of robots.
Then we will calculate the cost of manufacture of all S-
batches of the products Rz = Rz ( Rz1 Rz 2 ! Rzi ! Rzf ) , (24)
G = å Gk N k . (18) where Rzi is the type of robot performing an i-type TO at a
i
minimal cost (it can be any of the (f+1) types).
Finally, we will calculate the required number of workers:

Authorized licensed use limited to: University of Liverpool. Downloaded on July 26,2020 at 23:44:22 UTC from IEEE Xplore. Restrictions apply.
C. Calculation of the minimal product cost VI. CONCLUSIONS DESIGN OF AN AUTOMATED
The optimal cost of one k-type product is calculated using AUTONOMOUS PRODUCTION SYSTEM
the formula: Algorithm of organizing an automated autonomous
production system is as follows.
Ck = å zi nki ; k=1, 2,…, S. (25) 1). Determining the optimal automation vector that aligns
i with the existing system of robots and the given list of TOs. It
is necessary to consider two automation schemes: one of which
where zi is the cost of performance of i-type TOs by a Rzi will minimize the product manufacture time and the other one
robot from (24) that automatically leads to the minimal product will minimize the product cost.
cost achievable in this system of robots.
2). Calculating the cost of products when they are
The cost of all batches of products manufactured following the optimal scheme.
3). Determining the load volume for each type of robot.
C = å Ck N k . (26)
k 4). Determining the required number of workers.
5). Determining the required number of robots of each type.
The algorithm (22) guarantees labour saving if compared
with manual assembly DG = G - C ³ 0 . 6). Determining the optimal automation scheme for the
purchased robots.
D. . Working time of each type of robot in the system of robots
7). Determining the optimal automation scheme for the
With the next step we will define the number and types of produced robots.
robots to be purchased so that the whole plan of product
manufacture was carried out by robots. We will calculate what REFERENCES
part of the service life of each type of robot will be needed for [1] G. Eason, B. Noble, and I. N. Sneddon, “On certain integrals of
the optimal cost plan of product manufacture. Lipschitz-Hankel type involving products of Bessel functions,” Phil.
Trans. Roy. Soc. London, vol. A247, pp. 529–551, April 1955.
The time spent by a Rzi-type robot to manufacture all k- (references)
type products can be calculated with the formula [2] Vassilyev S.N., Novikov D.A., Bakhtadze N.N. Intelligent Control of
Industrial Processes, IFAC Conference on Manufacturing Modelling,
qki = N k nki bi , (27) Management and Control, 2013 pp. 49-57.
[3] Garnier S., Dumas C., Caro S., Furet B. Quality Certification and
Productivity Optimization in Robotic-based Manufacturing, IFAC
where Nk is the amount of a k-type product batch, nki is the Conference on Manufacturing Modelling, Management and Control,
number of i-type TOs needed to manufacture one k-type 2013, pp. 855-860.
product, biz is the time needed to perform i-type TOs by a Rzi [4] Wencai Wang and Yoram Koren. Design Principles of Scalable
Reconfigurable Manufacturing Systems, IFAC Conference on
robot performing this operation at a minimal cost. Manufacturing Modelling, Management and Control, 2013. pp. 1427-
The number of robots performing i-type TOs at a minimal 1432.
cost to carry out the whole plan is calculated using the formula: [5] Chadeev V.М., Aristova N.I. Self-reproduction of mechanical robots.
Moscow: SINTEG, 2012. 312 pp. ISBN 978-5-89638-123-5.
[6] Aristova N.I., Chadeev V.M. Estimation of the mass production
Qi
N Ri = automation effect. // Automation in Industry. 2016. №5.
TRi , (29) [7] Chadeev V.М., Aristova N.I. Control Of Industrial Automation, IEEE
Xplore Digital Library, Tenth International Conference Management of
Large-Scale System Development (MLSD). 2017.
where NRi is the number of robots, Qi is the scope of work [8] Chadeev V.М., Aristova N.I. Production Automation Control Theory
and Tri is the service life of a robot. Invariable with the Industry, IEEE Xplore Digital Library, Tenth
International Conference Management of Large-Scale System
Development (MLSD). 2018.

Authorized licensed use limited to: University of Liverpool. Downloaded on July 26,2020 at 23:44:22 UTC from IEEE Xplore. Restrictions apply.