You are on page 1of 5

2017 3rd International Conference on Control, Automation and Robotics

Safe Human-Robot-Collaboration-Introduction and Experiment Using ISO/TS


15066

Martin J. Rosenstrauch J6rg Kri.iger


Institute for Machine Tools and Factory Management Institute for Machine Tools and Factory Management
Technische Universitat Berlin Technische Universitat Berlin
Berlin, Germany Berlin, Germany
email: rosenstrauch@tu-berlin.de email: joerg.krueger@tu-berlin.de

Abstract-The use of industrial robots for human-robot­ more specific standard for industrial robots is EN ISO 10218
collaborative applications continually gains in significance. The [4]. These standards are basic guidelines for identifying
advantages of both human and robots can be combined to possible hazards, evaluate them and reduce the risk of
minimize the disadvantages. Human skills like cognition, accidents. Basically three main steps have to be repeated
adaptation or tactile abilities can be added with robot until the residual risk of eligible hazards is minimized or
strengths like speed, force or precision. Besides its advantages brought to an acceptable level.
,-----,
this development also cause new challenges regarding safety 1---------
measures, particularly to ensure humans integrity working 1
1 1 1 1
closed to or even with heavy robots. Therefore in February 1 1 1 1
2016 the International Organization for Standardization (ISO) 1 1 1 1
I 1 1 1
published a new specification ISO/TS 15066 (Robots and
1 1 1 1
robotic devices - Collaborative robots), a supplement to EN ISO 1 1
10218 (Robots and robotic devices - Safety requirements for
L _________I
industrial robots). To get a comprehensive overview of the new Figure I. From conventional safe guards to collaborative operation modes
specification this paper first gives an introduction into safety
standards and guidelines for risk assessment in general and The first step within a risk assessment is a risk analysis
industrial robots specifically, followed by a more detailed
which means the identification of all risks or hazards that
insight into ISO/TS 15066. An experiment with an exemplary
may occur during all kind of operation modes combined with
standard collaboration scenario demonstrates its usage. A
all possible activities (setting, process changeover, etc.).
subsequent discussion about the illustrated residual hazard
Unintended behaviour of the operator or unpredictable
potential in case of an incident despite compliance with ISO/TS
behaviour of the robot system must be taken into
15066 and a prospect on future approaches to achieve more
safety within human-robot-collaboration will complete this
consideration. Most typical hazards are mechanical ones like
paper.
bruising, cutting, clamping or impacts. Furthennore there can
be electrical or thermal hazards, or additional risks due to
Keywords-ISOITS 15066; safety; risk assessment; human­ noise, vibration, radiation, chemicals and many others.
robot-collaboration; shared workspace The risk analysis is followed by a risk evaluation of
tagged hazards. Here probabilities and extents of damages or
I. INTRODUCTION injuries are assessed. The results can be divided into different
Nowadays manufacturing moves from conventional static "Performance Level" (PL) or "Safety Integrity Level" (SIL).
production lines with large lot sizes to more flexible and PL is defined in EN ISO 13849-1 [5]. SIL is introduced in
easily adaptable factory units to manufacture or assemble EN 61508 [6] and EN 62061 [7]. The different level in PL
small batches or even individual constructions. In the course and SIL both are based on the probability of dangerous
of this process collaboration of humans and robots become failure per hour and can be transformed into each other
more and more important [1], [2]. The exclusive use of through some kind of lookup table.
industrial robots separated from workers by physical safe
guards was extended by operation modes, where workers act risk assessment
lEN ISO 1Z100:2010) risk analysis
very proximate to or even collaborative with robots (Fig. 1).
As a consequence existing standards and requirements
regarding safety are no longer sufficient. risk evaluation

A. Industrial Robot Safety and Risk Assessment


risk reduction
Industrial robots are large and powerful machines with
huge operating ranges and often with unpredictable and residual risk
unanticipated movements. For this reason a risk assessment
is required to put robots into operation. A general standard
for safety of machinery is given with EN ISO 12100 [3]. A Figure 2. ISOITS 12100:2010 basic procedure of risk assessment

978-1-5090-6088-7117/$31.00 ©2017 IEEE 740


After having identified and evaluated possible risks transmitting the manual input into robot motion
particular hazards are reduced by diverse protective (Fig. 3, c).
measures or safeguards like tripping functions, guards or • Power and force limiting finally allows a
others. As shown in Fig. 2 the procedure described is finally completely shared collaborative workspace and the
repeated iteratively with a risk comparison after each possibility of unintentional and unpredictable
iteration until the residual risk is reduced to minimum. collisions between human and robot. In order to
provide safety power and force are limited to ensure
B. Safety in Human-Robot-Collaboration compliance with given biomechanical force or
Earlier work like [8]-[10] already recognized the pressure thresholds (Fig. 3, d). These biomechanical
necessity of research in the area of safety. Only that allows a limits (maximum pressure and forces) for quasi­
transfer of scientific progress in the field of human-robot­ static and transient contact depend on different
collaboration to an industrial level. Besides technological body parts, shown as red dots in Fig. 4. Equations
approaches of safety improvement by using for example for calculating transient contact speed limit values
optical or capacitive sensors [11], [12] another research are introduced. Due to this, it is possible to derive
subject was the investigation of injury levels caused by a and implement concrete settings (maximum speed,
collision between human and robot [13]-[15]. force) to the robot system to ensure safety after risk
The established standard ISO 10218 is not sufficient to assessment.
archive both, maximum safety and a flexible human-robot­
collaboration usage. Knowing that for years [16] the
International Organisation of Standardization specified the
requirements for collaborative industrial robots and
published the technical specification ISO/TS 15066:2016
[17]. /
II. TECHNICAL SPECIFICATION ISO/TS 15066
The technical specification ISO/TS 15066 is not a
standard, it rather updates the state of the art in existing robot
safety standards. Now it is accepted as best practise in
addition to DIN EN 10218 regarding human-robot­ Figure 3. 1S0rrS 15066:2016 collaborative operation modes
collaboration. While DIN EN 10218 states that any contact
"should not create an injury", ISO/TS 15066 names limits for
force and speed to comply that. The specification does not
replace risk assessment given in 10218 but makes reasonable
for the first time. Because of that it is planned to add its
content to ISO EN 10218 in the next review in 2017. ISO/TS
15066 first introduces into general hazard identification and
risk assessment regarding to human-robot-collaborative
systems. After that requirements for collaborative robot
system applications are presented. A distinction is made here
between four different operating modes:
The four operation modes in more detail are: Figure 4. 1S0rrS 15066:2016 body model and different body parts
• Safety-rated monitored stop only allows a robot
motion if the operator is outside of the collaborative The reason for dividing human-robot-collaboration into
workspace. As soon as he enters to interact, the different operation modes is to be found in different
robot motion stops until he exits (Fig. 3, a). application areas. Cases of manufacturing, where power or
• Speed and separation monitoring goes one step force usually need to be fixed, can be handled by safety­
further by allowing the operator to enter the monitored steps while common use-cases like assembly or
collaborative workspace. Here the robot motion packaging are best covered by the introduced modes power
speed de- or increases dynamically depending on and force limiting or hand guiding.
the distance between
III. EXPERIMENT
operator and robot. If the distance gets below a
minimum protection separation distance the robot In this part an experiment is made to demonstrate the
stops (Fig. 3, b). usage of ISO/TS 15066 and also to demonstrate that a
• Hand guiding goes another step further. In this first human-robot system might be still hazardous if the risk
case of accepting a contact between human and assessment for a given application is inexpertly done. The
robot the operator is able to guide the robot by scenario contains a pick-and-place task, as shown on top in
moving it by hand within the collaborative Fig. 5. Within an abstract assembly process a work piece
workspace, e.g. by a hand-guiding device or by the symbolized by a cube shall be picked and moved (drawn
use of force torque sensor at the tool center point continuous line) to another position.

741
Fmax= 140N (1)

Pmax 200� (2)


cm2
These limits (given in [17]) are double in case of a
transient contact. The maximum relative speed between
human and robot can be calculated:
Fmax
V,'el-max (3)

=
Fmax
1nnrnl1 k (4)
mH+ml1.

m
0.6730 (5)
s

where mH shall be assumed with 0. 6kg for a human hand and


Figure 5. exemplary pick-and-place task (top) and unintended incident mR is the sum of half of the total mass of the moving parts of
caused by unpredicted and unintended human operation (bottom)
the robot and the effective payload, all together estimated to
15kg. k is called effective spring constant and for the back of
As it is planned to have a human-robot-collaboration
within that assembly process in general all guards are a hand given with 75 n�n. The maximum transfer energy E
reduced to a level where it is possible for a worker to place allowed during a transient contact is limited to 0,49J. The
something into or grasp something out of the robot operating transferred energy is related to the robot speed:
area. Due to that an unintended contact between human and 1
robot may occur (Fig. 5 bottom). Within a risk assessment E 2f..lVre1 (6)
the operation mode power and force limiting was selected.
=> Vrel 2f..lE (7)
A. Experimental Setup
The used kinematic is a standard 6 DOF robot ann
0.5654 m (8)
(consisting of Schunk Amtec PowerCube modules) with a s
gripper as end effector. As specific body area the back of a
workers hand is chosen. A potential hazard might be a The maximum speed Vmax is either is limited to its direct
transient contact between gripper and hand followed by calculation (9) or results out of the maximal allowed transfer
clamping and dragging along the hand while continuing the energy (6):
planned pick-and-place task. Forces applied on the hand by Vmax max( Vrel, Vrel-max ) (9)
the robot were measured with a force torque sensor system,
to ensure that they are within allowed limits. The complete
=> Vmax V"el 0.5654 m (10)
s
=

experimental setup can be seen in Fig. 6.


Summarised: under the tenns of the scenario outlined
above and after a done risk assessment a robot could possibly
clamp the back of a humans hand with a force of 140N on a
6 OOF robot -::==:!...; contact area of less than lcm2, e.g. with the front of one
single gripper jaw. That could happen in a trivial way while
the worker tries to grasp a work piece carelessly and his hand
gets between work piece and gripper. In this moment the
relative speed between human hand and robot gripper can be
quite less than the allowed Vmax � 0.59. Even though
Figure 6. Experimental setup ISO/TS 15066 was considered it is obvious that serious
injury might be the consequence. Nevertheless ISO/TS
As described above in this case risk reduction is achieved 15066 states: "the limits will prevent the occurrence of skin
by limiting the maximum force and torque of the robot or soft tissue penetrations".
system which in turn is effected by limiting power and speed. To demonstrate the assumption of a still hazardous
The required values are either given in ISO/TS 15066 [17] or situation, although ISO/TS 15066 was applied, the
determined by equations also given in [17] accordingly. The introduced pick-and-place task with a simulated quasi-static
maximum permissible pressure and force for this particular contact is done with a piece of pork belly, which is for
body area and the case of a quasi-static contact has to be example often used in medical teaching to simulate human
limited: back skin [18].

742
B. Experimental Results ISO/TS 15066 was_applied in a reasonable way the human
The experiment confirmed the assumption made before. skin simulating pork belly was seriously violated. As it is
In particular the safety limit for Fmax calculated in accordance written within the specification itself, limits for force and
with ISO/TS 15066 led to considerable strains to the pork pressure are taken out of a single study. The given values are
belly. Fig. 7 presents the smoothed forces in X-, y- and z­ individual pain thresholds which might cause painful
direction measured by a force torque sensor. Besides a accidents with substantial injuries. The calculation procedure
clearly quantifiable impact in z-direction (Fig. 7 top left) a for power and force limiting is based on modem lightweight
horizontal movement (Fig. 7 top right) causes additional designed robots; widely used heavy industrial robots are not
forces along the direction of movement, which in turn would addressed. Furthermore "objects with sharp, pointed,
cause violations between layers of tissue. shearing or cutting edges, such as needles, shears, or knives,
r-------, and parts which could cause injury shall not be present in
the contact area" [17]. Those facts limit the usability in
addition. Unknown factors such as Vhuman are still difficult to
handle and cannot be fully covered. In this case the challenge
is to assess that unknown which is an important factor for the
resulting impulse between human and robot in cause of a
collision, because even if the robot speed is very limited,
human could move or accelerate that fast [19], [20], that the
resulting energy could be higher than permitted. Not yet
widely researched is the general acceptance of a human­
robot-shared workspace. Studies about fundamental
questions like how heavy or fast a robot might be to be
accepted in a shared workspace by a human worker or
80 aspects like the predictability of the robot movement with
varying paths due to collision avoiding path planning or
�60 varying tasks are still missing.

[�:
fl40 A. Future Work
.E 20 F
z Firstly it seems expedient to pursue the identification and
--

0 ...;.
division of different application scenarios and to specify
0 2 3 5 6 concrete and more individual risk assessment guidelines.
time [51 Therefore limits for force and speed are expected to evolve
by investigating body zone dependent pain thresholds in
Figure 7. Measured forces in X-, y- and z-direction future studies. On this basis it is to be expected that new
permitted limits for force and speed limits will be lower as at
The pork belly skin was put under heavy strain. Minor present. As a consequence flexible human robot
cuts were added. To get a rough idea of the measured impact collaboration could become be restricted. As well it seems
fig. 8 shows the depth of penetration (approx. 25mm) at a appropriate to identify existing technologies, approaches and
contact of only one gripper jaw with Fz = 85N, which is methods and further use them to confirm the adherence of
about 60% of the allowed force for a clamping contact. latest safety standards. Thus, for example, speed limits of
Completely not taken into account so far is the risk of bone robot movement and possible resulting impulse energies
injury which is in this context hard to simulate, especially could be monitored by optical systems or acceleration
metacarpals. sensors. Such approaches could also support the
improvement of workers acceptance in a safe human-robot­
collaboration. In addition it is conceivable to combine
different safety systems in a redundant way. Current research
in cloud based service technologies with abstract infinite
computing power only limited by transfer rates and latencies
allow new approaches in merging various safety systems.

V. CONCLUSION

The most important statement of ISO/TS 15066 is an


implicit one: a safe robot does not exist, at best safe
Figure 8. quasi-static contact - depth of penetration with a force F,;::O 85N
applications. But safety here means that there is still a risk of
serious injury. To assume that an industrial robot is safe as
IV. DISCUSSION
long as the specification is taken into account to ISO 10218
The experiment done illustrates the decreased but still does not comply to an expertly application. The present
existing gap between a feasible guideline for safety in human paper briefly introduced into industrial robot safety, risk
robot collaboration and the resulting real safety. Even though assessment and especially the current state of human-robot-

743
collaboration safety. The growing trend towards shared Manufacturing (ISAM), 2011 iEEE international Symposium on, May
workplaces caused the development of a new specification 2011,pp. 1-6.

refining safety requirements - called ISO/TS 15066. For a [9] M. Giuliani,C. Lenz, T. Muller, M. Rickert, and A. Knoll, "Design"
principles for safety in human-robot interaction," international
better understanding of the specifications significance and to Journal of Social Robotics, vol. 2,no. 3,pp. 253-274,2010.
allow a correct apply its content was explained in more detail
[10] J. T. C. Tan,F. Duan,Y. Zhang,R. Kato,and T. Arai,"Safety design
supplemented by an experimental example use case. A fmal and development of human-robot collaboration in cellular
discussion emphasized possibilities and limits of the latest manufacturing," in 2009 IEEE International Conference on
international guideline on safety regarding human-robot­ Automation Science and Engineering, Aug 2009,pp. 537-542.
collaboration and gave a prospect on further steps towards [II] c. Morato, K. N. Kaipa, B. Zhao, and S. K. Gupta, 'Toward safe
achieving a maximum in human robot safety within human robot collaboration by using multiple kinects based real-time
human tracking," Journal of Computing and information Science in
changing requirements.
Engineering, vol. 14,no. I,p. 011006,2014.

REFERENCES [12] B. Sekoranja, D. Ba' si' c, M. ' Svaco, F.' Suligoj, and B. Jerbi' c,
"Human-' robot interaction based on use of capacitive sensors,"
[I] 1. Krueger, T. Lien, and A. VerI, "Cooperation of human and Procedia Engineering, vol. 69,pp. 464-468, 2014.
machines in assembly lines," {CiRP} Annals - Manufacturing
[13] S. Haddadin,A. Albu-Schaffer,and G. Hirzinger,"Requirements for"
Technology, vol. 58,no. 2,pp. 628 - 646,2009.
safe robots: Measurements, analysis and new insights," The
[2] 1. Fryman and B. Matthias, "Safety of industrial robots: From international Journal of Robotics Research, vol. 28, no. 11-12, pp.
conventional to collaborative applications," in Robotics; Proceedings 1507-1527,2009.
of ROBOTiK 2012; 7th German Conference on, May 2012,pp. 1-5.
[14] S. Haddadin, S. Haddadin, A. Khoury, T. Rokahr, S. Parusel, R.
[3] ISO 12100:2011-03, "Safety of machinery - general principles for Burgkart, A. Bicchi, and A. Albu-Schffer, "A truly safely moving
design - risk assessment and risk reduction (ISO 12100:20I0)," robot has to know what injury it may cause," in 2012 IEEE/RSJ
International Organization for Standardization,Standard DIN EN ISO international Conference on intelligent Robots and Systems, Oct 2012,
12100,Mar. 2012. pp. 5406- 5413.
[4] ISO 10218-1:2012-01, "Robots and robotic devices - safety [15] C. A. Cordero,G. Carbone,M. Ceccarelli,J. Echvarri,and 1. L. Muoz,
requirements for industrial robots - part I: Robots (ISO 10218- "Experimental tests in human-robot collision evaluation and
1:2011)," International Organization for Standardization, Standard characterization of a new safety index for robot operation,"
DIN EN ISO 10218-1,Jan. 2012. Mechanism and Machine Theory, vol. 80,pp. 184 - 199,2014.
[5] ISO 13849-1:2015, "Safety of machinery - safety-related parts of [16] 1. Fryman, "Updating the industrial robot safety standard," in
control systems - part I: General principles for design (ISO iSRIRobotik 2014; 4ist international Symposium on RobotiCS;
138491:2015)," International Organization for Standardization, Proceedings of, June 2014,pp. 1-4.
Standard DIN EN ISO 13849-1,Jun. 2015.
[17] ISO/TS 15066:2016, "Robots and robotic devices - collaborative
[6] IEC 61508-1:2010, "Functional safety of robots," International Organization for Standardization, Standard
electrical/electronic/programmable electronic safety-related systems ISO/TS 15066:2016,Feb. 2016.
part I: General requirements (TEC 61508-1:2010)," DIN German
[18] X. Wang, Y. Albahrani, M. Pan, and J. Levitt, "Skin simulators for
Institute for Standardization,Standard DIN EN 61508-1,Feb. 2016.
dermatological procedures," Dermatology online journal, vol. 21,no.
[7] IEC 62061:2005 + AI:2012 + A2:2015), "Safety of machinery - 11,2015.
functional safety of safety-related electrical, electronic and
[19] S. Kajikawa,M. Saito, K. Ohba, and H. Inooka, "Analysis of human
programmable electronic control systems (IEC 62061:2005 + al:2012
arm movement for catching a moving object," in Systems, Man, and
+ a2:2015)," DIN German Institute for Standardization,Standard DIN
Cybernetics, 1999. iEEE SMC '99 Conference Proceedings. 1999
EN 62061,May 2016.
iEEE international Conference on, vol. 2,1999,pp. 698-703 vol.2.
[8] B. Matthias, S. Kock, H. Jerregard, M. Kilman, and 1. Lundberg,
[20] N. Fligge, J. Mcintyre, and P. van der Smagt, "Minimum jerk for
"Safety of collaborative industrial robots: Certification possibilities
human catching movements in 3d," in 2012 4th iEEE RAS EMBS
for a collaborative assembly robot concept," in Assembly and
international Conference on Biomedical Robotics and
Biomechatronics (BioRob), June 2012,pp. 581-586.

744

You might also like