Introduction

• Introduction to robots
• Industrial Robots
• Service Robots
• Robot components
• Actuators
• Sensors

Prepared by: Tayachew Fikire

Robotics(EMEg4242) 1
 What is a robot?
 industrial definition (RIA = Robotic Institute of America)
re-programmable multi-functional manipulator
designed to move materials, parts, tools, or specialized devices through
variable programmed motions for the performance of a variety of tasks,
which also acquire information from the environment
and move intelligently in response

 ISO 8373:2012 definition

an automatically controlled, reprogrammable, multipurpose manipulator
programmable in three or more axes, which may be either fixed in place or
mobile for use in industrial automation applications

 more “visionary” definition

intelligent connection between perception and action

Robotics 2
 Robot types
 Robot Manipulators
Start

 Mobile Robots

3
 Mobile Robots
 Walking Robots
Start

 Humanoid Robots

4
 Mobile Robots

Legged Robot
Start Wheeled Robot

5
 Mobile Robots

Autonomous Underwater
Start Vehicle Unmanned Aerial Vehicle

6
 What Can Robots do?

Start

Jobs that are dangerous

for humans

Decontaminating Robot
Cleaning the main circulating pump housing
in the nuclear power plant

7
 What Can Robots do?

Start

Repetitive jobs that are

boring, stressful, or labor-
intensive for humans

Welding Robot

8
 What Can Robots do?

Start

Manual tasks that human

don’t want to do

The SCRUBMATE Robot

9
 Robots Manipulator anatomy

 Manipulator consists of joints and links Link3

Start Joint3
 Joints provide relative motion
End of Arm
 Links are rigid members between joints
 Various joint types: linear and rotary
 Each joint provides a “degree-of-
Link2
freedom”
 Most robots possess five or six degrees- Link1
of-freedom
 Robot manipulator consists of two sections: Joint2

 Body-and-arm – for positioning of

Joint1
objects in the robot's work volume
 Wrist assembly – for orientation of Link0
objects Base

10
 Manipulator Joints
 Translational motion L
Start
 Linear joint (type L):
O
 Orthogonal joint (type O)

R
 Rotary motion
 Rotational joint (type R)
 Twisting joint (type T)
 Revolving joint (type V)
T

• Axis of rotation translation V

• Axis of input and out put links

11
 Joints notations
 Uses the joint symbols (L, O, R, T, V) to designate joint types
used to construct
Start robot manipulator
 Separates body-and-arm assembly from wrist assembly using
a colon (:)

 Example: TLR : TR

 Common body-and-arm configurations …

12
 Body & arm configuration

 Notation TRL:
Start

 Consists of a sliding arm (L joint) actuated relative to the body,

which can rotate about both a vertical axis (T joint) and horizontal
axis (R joint)

13
 Body & arm configuration

 Notation LOO:
Start

 Consists of three sliding joints,

two of which are orthogonal
 Other names include rectilinear
robot and x-y-z robot

14
 Body & arm configuration
 Notation TRR:
Start

15
 Body & arm configuration
Notation VRO
Start

Sketch following manipulator configurations

(a) TRT:R, (b) TVR:TR, (c) RR:T

16
 Wrist configuration
 Wrist assembly is attached to end-of-arm
Start
 End effector is attached to wrist assembly

 Function of wrist assembly is to orient end effector

 Body-and-arm determines global position of end effector
 Two or three degrees of freedom:
 Roll
 Pitch
 Yaw

 Notation :RRT

17
 End effectors

 The special tooling

Start for a robot that enables it to
perform a specific task
 Two types:
 Grippers – to grasp and manipulate objects (e.g.,
parts) during work cycle
 Tools – to perform a process, e.g., spot welding,
spray painting, drilling ,cutting

18
 Grippers & Tools

Start

19
 Robot hands

• the UBHand series of robot dexterous hands (Univ Bologna)

3 fingers with 9 degrees of freedom, tendon actuation,
supporting palm, and tactile sensors on all phalanges

Robotics
20
 Anthropomorphic UBHand IV

data glove
for motion
capturing

• the UBHand IV has deformable elements as joint hinges (compliant mechanisms);

the endo-skeletal structure with 5 fingers may host distributed sensors
and continuous compliant cover (G. Vassura, Univ Bologna)

Robotics
21
 Integration of robot hand + arm

• the complete UBHand II, with electrical motors

and electronics presented at EXPO92 in Seville

•integration in the forearm

of the Unimation PUMA 560

Robotics 1 22
 Anthropomorphic upper limbs

• Justin robot has 7+7+3 degrees of • the robot developed in the

freedom + many dofs in the two hands German national project
(DLR, Germany) on humanoids
Robotics
23
 Workspace

Start

Model categories

24
 Workspace

Robotics 25
 Robot Programming
 Leadthrough programming
Start
 Work cycle is taught to robot by moving the manipulator through the
required motion cycle and simultaneously entering the program into
controller memory for later playback
 Robot programming languages
 Textual
Model programming
categories language to enter commands into robot controller
 Simulation and off-line programming
 Program is prepared at a remote computer terminal and downloaded to
robot controller for execution without need for lead through methods.

26
 Industrial and
service Robotics

Robotics 27
 Robots !!

Spirit Rover
(2002)
Comau H4 Waseda WAM-8
(1995) (1984)

Robotics 1 28
 Evolution toward industrial robots
computer 1950 mechanical
numerically controlled
telemanipulators
(CNC) machines

robot
manipulators

1970
Unimation PUMA

 with respect to the ancestors

 flexibility of use
 adaptability to a priori unknown conditions
 accuracy in positioning
 repeatability of operation

Robotics 1 29
 The first industrial robot

US Patent General Motor plant, 1961

G. Devol and J. Engelberger (Unimation)

Robotics 1 30
 Robot manipulators

ASEA IRB-6 Hirata AR-300

(1973) (1978)
first robot first SCARA
all-electric-drives robot

Cincinnati
Unimation
Milacron T3
PUMA 560
(1974)
(1979)
first micro-
6R with
computer
human-like
controlled dexterity
robot

Robotics 31
 Industrial & service robots
Industrial robots
 automatically controlled, programmable, multipurpose, 3+ axes
 for use in industrial automation applications.
 equipped with application-specific end-effectors.
Service robots
 perform tasks excluding industrial automation
 usually application-specific design, often fewer than 3 axes
 sometimes not fully autonomous but remote-controlled

different customers, pricing, machinery, distribution channels, suppliers

… but separation
line is blurring:
same unit can act
as both, depending
on the application
Robotics 32
 Industrial applications
• manipulation (pick-and-place, handling,
machine feeding)
• assembly and packaging
• spray painting and coating (nozzles)
• arc welding or spot welding (with
pneumatic or servo-controlled guns)
• laser cutting and welding
• gluing and sealing
• mechanical machining operations
(milling, drilling, deburring, grinding, …)

Robotics 33
 Robot manipulator kinematics

Kuka 150_2 S2000 Comau Fanuc

open kinematic chain Smart H4 F-200iB
(series of rigid bodies closed kinematic chain parallel kinematics
connected by joints)

Robotics 34
 SCARA-type robots

Mitsubishi RP Mitsubishi RH
(repeatability 5 micron, (workspace 850 mm, Bosch Turbo
payload 5 kg) velocity 5 m/s)

SCARA (Selective Compliant Arm for Robotic Assembly)

• 4 degrees of freedom (= joints): 3 revolute + 1 prismatic (vertical) axes
• compliant in horizontal plane for micro-assembly and pick-and-place

Robotics 35
 Other types of robots

Comau Mast ABB Flexpicker

gantry robot (150 pick-and-place
(payload up to 560 kg) operations/minute)

Robotics 36
 Mobile base robots in industry

• AGV (Automated Guidance Vehicles) for material and parts transfer

on the factory floor: wire- or laser-driven along predefined paths

Robotics 37
 service Robots

Robotics 38
 Some application domains

 extreme environments  mine exploration

 space  de-mining
underwater

 civil and naval construction
 medical robotics
 automatic refueling
 assistive
 rehabilitative  museum guide
 surgical  fire fighting
 home cleaning  inspection and surveillance
 agriculture  emergency rescue
 lawn mowing  entertainment
 food industry  humanoids

professional & personal service robots

Robotics
39
 Service robots on the market!

Bluebotics Esatroll - Paquito 2.0 Yujin GoCart2

logistics in factory floor elderly and health care Cyberdyne HAL
exoskeleton for walking

Vorwerk Thymio educational

vacuum cleaner mobile robot
Lely Vector automated feeding
Robotics
40
 Space robotics

• DLR Rotex robot arm in a set of

• NASA Sojourner, first robot
experiments of the Spacelab-D2
to explore Mars in 1997
mission on the Columbia shuttle
in 1993

Robotics
41
 Space robotics
• robotic arm SSRMS (Canadarm) in
operation on the Space Shuttle, with
outstretch of about 17 meters

• service manipulator SPDM (Dextre)

on the International Space Station
(ISS), mounted on a supporting rail
and used to refuel satellites

Robotics
42
 Underwater robotics

• typically actuated by thrusters

(directional forces on the tail)
• cannot translate sideways
(“maneuvers” are necessary)

• Odyssey-IV (MIT)

• Odin-III, omni-directional • ROMEO in Antarctica (CNR,

(University of Hawaii) Automazione Navale, Genova)
Robotics
43
 Underwater robotics

Ansaldo underwater arm

performing a cable hooking task
(SAUVIM project)

UBC Gavia underwater robot

(University of British Columbia)

44
 Underwater robotics

• Aqua robot, amphibious robotic vehicle (McGill University)

• size and weight: 50x65x13 cm, 18 kg

• locomotion: through six independently actuated flippers
• maximum depth: 37 m
• sensors: two cameras (front/back), acoustic sensor for localization
(sonar), tri-ocular sensor (structured light)
• power source: 48V lithium battery

Robotics
45
 De-mining

• teleoperated mobile robot on • PEMEX lightweight anti-personnel

tracks used by the police for mine detector (EPFL, Lausanne)
bomb disposal • weight: 16 kg, max 6 kg for wheel
• two 70 W DC motors (vel 2 m/s)
• oscillating sensorized head

Robotics 1 46
 Medical robotics
patient aid

• MOVAID project for the aid of disabled people

in home activities (Scuola Sup Sant’Anna, Pisa)

• deambulation support system

PAM-AID (Trinity College, Ireland)

Robotics
47
 Medical Robotics
rehabilitative

• robotic arm with shoulder and elbow having full mobility and with a
gripper hand (Pittsburgh University)
• in tests on monkeys (with immobilized upper limbs), motion commands
sent to the arm by the central nervous system (brain) are measured by
a set of electrodes and used to command the robotic arm
Robotics
48
 Medical robotics
rehabilitative

• commercialized by Ossur (Iceland)

• a prosthesis sensorized at the knee (angle

and force), capable of processing sensor
data and of extracting a gait model of the
user, so as to adapt its dynamical behavior
(knee motion and stiffness)

Robotics
49
 Medical robotics
rehabilitative

• "RUPERT" Robotic Upper Extremity Repetitive Therapy (Arizona State

University + Kinetic Muscles, Inc.)
• sustains the human arm with pneumatic muscles (McKibben actuators)
• it can be programmed for the execution of cyclic exercises of rehabilitation

Robotics
50
 Surgical robotics

command station

overview of the operating room

(haptic) interface
• da Vinci© system (Intuitive Surgical
Inc.) [see the course “Medical
Robotics Robotics”] 21
 Surgical robotics

• Robodoc by Integrated Surgical Systems (USA)

was used first for orthopedic surgery (ankle replacement)

• Steady-Hand force-assisted system (Johns Hopkins Univ)

improves accuracy and repeatability of surgeons allowing
task-driven compliance
Robotics
52
 Home cleaning

• vacuum cleaner robot • commercial video

Trilobite by Electrolux (Sweden) iRobot Roomba 560 (USA)
‐ n o w available everywhere!
Robotics
53
 Cleaning robot contest

•competition among robot vacuum cleaners

in home environments (IROS’02, Lausanne)
Robotics
54
 Cleaning of external surfaces

• Skywash cleans civil airplane bodies • a robot prototype for cleaning large
and is “the largest robot worldwide” glass windows of civil buildings
(AEG/Dornier/FhG-IPA/Putzmeister)

Robotics
55
 Lawn mowers

• Automower autonomous robot by Husqvarna (Sweden)

has low power consumption (biocut) and solar recharge

Robotics
56
 Food industry

• AdeptOne SCARA robot with

• Ulixes robot by IMT (Germany) 4-sausage gripper
aligns 10000 sausages per hour
Robotics
57
 Automatic refueling

• cooperation of Reis Robotics,

Mercedes, BMW, and IPA Fraunhofer

• Smart Pump system (USA)

Robotics
58
 Mine exploration

• Groundhog (Carnegie Mellon)

• 750 kg, double axes, articulated
• movable SICK laser (rangefinder)
• gas and immersion sensors
• SLAM algorithm (Simultaneous
Localization And Mapping)

Robotics
59
 RoboCup and RoboRescue

•RoboCup middle-size league • Orpheus robot won the RoboRescue

(wheeled mobile robots, here (exploration and search of victims
with omni-directional vision) in a disaster environment)

2003 edition, Padova Fair

Robotics
60
 DARPA Grand Challenge

the “Ghostrider” motorcycle

5 SICK lasers for mapping and localization testing in Nevada
on the 2005 winning VW Touareg “Stanley”

• competition for fully autonomous vehicles on a long mixed-type track

Robotics
61
 Robots filling a glass of beer

two cooperating ABB robots

a single KUKA robot

Robotics
62
 Museum guidance

• three mobile robots for museum guidance

(Museum für Kommunikation, Berlin)
Robotics
63
 Entertainment

• the Anaconda robot (Edge Innovations, USA) weights various tons,

has 60 artificial spinal vertebrae, is 12 meters long,
and is actuated by hydraulic motors so as to reach a speed of up to 60 km/h
Robotics
64
 Human motion replication

• the anthropomorphic robot by Sarcos Entertainment Systems (USA)

replicates the movements of a human wearing a sensorized exoskeleton

Robotics
65
 Human-Robot Interaction

• physical and cognitive interaction between a Sarcos robot and a human

intrinsic mechanical compliance in the robot structure
is here more important than accuracy in motion execution

Robotics
66
 Human-Robot Interaction (HRI)

cognitive interaction (cHRI) physical interaction (pHRI)

in Robot@CWE EU Project in PHRIENDS EU Project

Robotics
67
 Human-robot collaboration

• Mr. Helper (Tohoku Univ) • CoBot scooter-like robot

collaborates in carrying heavy for mounting car doors
and/or large loads (General Motors)

Robotics
68
 Biomorphic robots

• biomorphic robots by MIT Leg Lab, USA:

Troody dinosaur and Flamingo bird

Robotics
69
 Humanoid robots

• the ASIMO project by Honda started in 1986

70
 ASIMO in action

ASIMO first and

climbing stairs second series
(Robodex 2003) (smaller size)

71
 Humanoid robots

• Sony SDR-3X
• HRP-2 (about 60 cm)
• humanoid robot
(58 kg, 150 cm, 30 dofs) (Q. Huang, PR China)
2002 Tokyo Univ

72
 Sony Q-RIO

group dancing video (2003)

• Sony Q-RIO
the first robot able to balance on a surf and stand up from the floor
(dead in 2006…)

Robotics
73
 Humanoid robots

height = 57 cm kinematic
weight = 4.5 kg description

• NAO, Aldebaran Robotics

since 2008, replaces AIBO quadrupeds in RoboCup standard league

Robotics
74
Robot components:
Actuators

75
 Robot as a system

commands actions
program working
Robot
of tasks environment

mechanical supervision
units units
sensor actuation
units units

76
 Functional units of a robot

 mechanical units (robot arms)

 serial manipulators: rigid links connected via rotational or prismatic
joints (each giving 1 degree of freedom = DOF)
 supporting structure (mobility), wrist (dexterity), end-effector (for task
execution, e.g., manipulation)
 actuation units
 motors (electrical, hydraulic, pneumatic) and transmissions
 motion control algorithms
 sensor units
 proprioceptive (internal robot state: position and velocity of the joints)
 exteroceptive (external world: force and proximity, vision, …)
 supervision units
 task planning and control
 artificial intelligence and reasoning

77
 Actuation systems

P = types of
power powers in play
supply
electrical, hydraulic,
or pneumatic mechanical
Pp
Pc Pa Pm Pu
power transmission
servomotor (mechanical gears)
amplifier
Pda Pds Pdt

electrical
power losses due to dissipative effects (e.g., friction)
power = voltage · current = pressure · flow rate = force · speed = torque · angular speed [W, Nm/s]
efficiency = power out/power in [%] energy ~ work = power · time [kWh, Nm, J]
78
 Desired characteristics
for robot servomotors

 low inertia
 high power-to-weight ratio
 high acceleration capabilities
 variable motion regime, with several stops and inversions
 large range of operational velocities
 1 to 2000 rpm (round per min)
 high accuracy in positioning
 at least 1/1000 of a turn
 low torque ripple
 continuous rotation at low speed
 power: 10 W to 10 kW

79
 Servomotors

 pneumatic: pneumatic energy (compressor) 

pistons or chambers  mechanical energy
 difficult to control accurately (change of fluid
compressibility)  no trajectory control
 used for opening/closing grippers
 ... or as artificial muscles (McKibben actuators)
 hydraulic: hydraulic energy (accumulation tank)
 pumps/valves  mechanical energy
 advantages: no static overheating, self-lubricated,
inherently safe (no sparks), excellent power-to-weight
ratio, large torques at low velocity (w/o reduction)
 disadvantages: needs hydraulic supply, large size, linear
motion only, low power conversion efficiency, high cost,
increased maintenance (oil leaking)
80
 Electrical servomotors

 advantages
 power supply available everywhere
 low cost
 large variety of products
 high power conversion efficiency
 easy maintenance
 no pollution in working environment
 disadvantages
 overheating in static conditions (in the presence of gravity)
 use of emergency brakes
 need special protection in flammable environments
 some advanced models require more complex control laws

81
 Electrical servomotors for robots
stator
stator
(permanent magnets)
collector

brushes

rotor (main
motor inertia) V1 V2 Vn
 switching circuit

armature
circuit Va Va

direct current (DC) motor with electronic switches (brushless)

82
 Advantages of brushless motors

 reduced losses, both electrical (due to tension drops

at the collector-brushes contacts) and mechanical
(friction)
 reduced maintenance (no substitution of brushes)
 easier heat dissipation
 more compact rotor (less inertia and smaller
dimensions)
… but indeed a higher cost!

83
 Principle of operation of a DC motor
permanent
magnets N-S
single coil
(armature)

commutator ring
(to switch direction
of armature current
DC supply every half round)
𝑣𝑎
𝐹⃗ = 𝐿 𝚤⃗× 𝐵 𝑟 = 𝑑 𝐹⃗

84
 DC electrical motor
mathematical model (in the time domain)

in absence of losses, conservation of

electrical balance power holds in energy transformation
(on the equivalent armature circuit)
𝑃𝑒𝑙 𝑒𝑐 = 𝑣𝑒𝑚 7 𝑖𝑎 = 𝑟𝑚 𝜔 = 𝑃𝑚 𝑒𝑐 𝑐
𝑑2𝑎 𝑡
𝑣𝑎 𝑡 = 𝑅 𝑎 𝑖 𝑎 𝑡 + 𝐿𝑎 + 𝑣𝑒𝑚 7 (𝑡) ⟹ 𝑘𝑣 = 𝑘 𝑡 (in SI units)
𝑑
𝑡
𝑣 𝑒𝑚7 𝑡 = 𝑘𝑣𝜔(𝑡) + 𝑖𝑎 𝑡 𝑅𝑎
𝐿𝑎
(back emf)
+ 𝑟 𝑚(𝑡) 𝑟𝑙 𝑜 𝑎 𝑑 (𝑡)
𝑣𝑎 𝑡
𝑣𝑒 𝑚 7 (𝑡) 𝐼𝑚

mechanical balance − 𝜔𝑡 𝐹𝑚
(Newton law on torques)
−
= 𝜃̇(𝑡)

𝑑𝜔 𝑡 using Laplace transform, differential

𝑟 𝑚 𝑡 = 𝐼𝑚 𝑡 + 𝐹𝑚 𝜔(𝑡) + 𝑟𝑙 𝑜 𝑎 𝑑 (𝑡)
𝑑 equations become algebraic relations!
𝑡 J
𝑟𝑚 𝑡 = 𝑘 𝑡 𝑖 𝑎 (𝑡)
𝑋 𝑠 = 𝑓 𝑥(𝑡) = H 𝑥 𝑡 𝑒 L𝑠𝑡 𝑑𝑡
(motor torque) I
85
 DC electrical motor
mathematical model for command and control

electrical balance Laplace domain mechanical balance

(transfer functions)
Va = (Ra + sLa) Ia + Vemf Tm = (sIm + Fm)  + Tload

Vemf = kv  Tm = kt Ia
current loop
kv = kt
ki
Ia Tload
V’c − Vc Gv Va 1
Tm - 1  1 
Ci(s) kt
1+sTv + sLa + sIm s
+ - - -

ki = 0  velocity generator* Ra Fm
ki Ci(0) Gv ≫ Ra  torque generator*
Vemf
* = the motor is seen here as a steady state “generator”
kv
in order to actually regulate velocity or torque in an DC motor
efficient way against Tload , further control loops are needed!

86
 Motion transmission

 optimize the transfer of mechanical torque from actuating

motors to driven links
 quantitative transformation (from low torque/high velocity
to high torque/low velocity)
 qualitative transformation (e.g., from rotational motion of
an electrical motor to a linear motion of a link along the
axis of a prismatic joint)
 allow improvement of static and dynamic performance by
reducing the weight of the actual robot structure in motion
(locating the motors remotely, closer to the robot base)

87
 Transmissions in robotics
 spur gears: modify direction and/or translate axis
of (rotational or translational) motor displacement
 problems: deformations, backlash
 lead screws, worm gearing: convert rotational
into translational motion (prismatic joints)
 problems: friction, elasticity, backlash
 toothed belts and chains: dislocate the motor
w.r.t. the joint axis
 problems: compliance (belts) or vibrations
induced by larger mass at high speed (chains)
 harmonic drives: compact, in-line, power efficient,
with high reduction ratio (up to 150-200:1)
 problems: elasticity
 transmission shafts: long, inside the links,
with flexible couplings for alignment
88
 Transmission gears in motion

 racks and pinion

 one rack moving (or both)

 epi-cycloidal gear train

 or hypo-cycloidal (small gear inside)

 planetary gear set

 one of three components is locked:
sun gear, planet carrier, ring gear

Robotics 1 89
 Optimal choice of reduction ratio
Pm Pu
transmission
motor link
gear

ideal case (no friction) Pdt power dissipated

. . by friction
Pm = Tm m = Tu u = Pu
torque × angular speed
. .
n = reduction ratio (≫1) m = n u Tu = n Tm
.. ..
to have u = a (thus m = n a), the motor should provide a torque
.. ..
Tm = Jm m + 1/n (Ju u) = (Jm n + Ju /n) a
inertia × angular acceleration

for minimizing Tm, we set: Tm = (J - J /n2) a = 0

m u
n
n = (Ju / Jm)1/2 “matching” condition between inertias

90
 Transmissions in industrial robots
 transmissions used (inside) 6-dof Unimation industrial robots with serial kinematics
PUMA 560

PUMA 260: 1st axis PUMA 560: 2nd and 3rd axes

PUMA 560: inner and outer links PUMA 560: last 3 axes
Robotics 1 25
 Inside views on joint axes 4, 5 & 6 of
an industrial KUKA robot

 looking inside the forearm to see the

transmissions of the spherical wrist
 motor rotation seen from the
encoder side (small couplings exist)

26
Robot components(Sensors ):
Proprioceptive sensors

9
3
 Properties of measurement systems - 1

 accuracy
agreement of measured values with a given reference
standard (e.g., ideal characteristics)
 repeatability
capability of reproducing as output similar measured
values over consecutive measurements of the same
constant input quantity
 stability
capability of keeping the same measuring characteristics
over time/temperature (similar to accuracy, but in the long run)

9
4
 Accuracy and Repeatability

average value measurements

* * ****
* *
* ****
*

target value
(ground truth)

low accuracy low accuracy high accuracy

low repeatability high repeatability high repeatability

better components! calibration!

3
 Accuracy and Repeatability
in robotics
 accuracy is how close a robot can come to a given point in its workspace
 depends on machining accuracy in construction/assembly of the robot, flexibility effects
of the links, gear backlash, payload changes, round-off errors in control computations, ...
 can be improved by (kinematic) calibration
 repeatability is how close a robot can return to a previously taught point
 depends only the robot controller/measurement resolution

 both may vary in different areas of

the robot workspace
 standard ISO 9283 defines conditions
for assessing robot performance
 limited to static situations (recently,
interest also in dynamic motion)
 robot manufacturers usually provide
only data on “repeatability”

simple test on repeatability of a

Fanuc ArcMate100i robot (1.3 m reach)
9
6
 Properties of measurement systems - 2

 linearity error
maximum deviation of the measured output from the
straight line that best fits the real characteristics
 as % of the output (measurement) range
 offset error
value of the measured output for zero input
 sometimes not zero after an operation cycle, due to hysteresis
 resolution error
maximum variation of the input quantity producing no
variation of the measured output
 in absolute value or in % of the input range

9
7
 Linearity, Offset, Resolution
Y
Ym

X = resolution

Y
output range

Ym = linearity error

Y0 = offset
X
Xm
input range 6
 Sensor measurements
some non-idealities

Asymmetry Bias Dead zone

Nonlinearity Scaling factor Quantization

99
 Classes of sensors for robots

 proprioceptive sensors measure the internal state of the

robot (position and velocity of joints, but also torque at joints or
acceleration of links)
 kinematic calibration, identification of dynamic parameters, control
 exteroceptive sensors measure/characterize robot
interaction with the environment, enhancing its autonomy
(forces/torques, proximity, vision, but also sensors for sound, smoke,
humidity, …)
 control of interaction with the environment, obstacle avoidance in

the workspace, presence of objects to be grasped, …

 mobile-base robots: localization in a map, navigation in unknown

environments, …

10
0
 Position sensors

 provide an electrical signal proportional to the displacement

(linear or angular) of a mechanical part with respect to a
reference position
 linear displacements: potentiometers, linear variable-
differential transformers (LVDT), inductosyns
 angular displacements: potentiometers, resolvers, syncros
(all analog devices with A/D conversion), optical encoders
(digital), Hall sensors, ...

the most used in robotics, since also linear

displacements are obtained through rotating
motors and suitable transmissions

10
1
 Absolute encoders
Photo-emitter
• rotating optical disk, with alternate
Optical disc transparent and opaque sectors on
multiple concentric tracks
• (infrared) light beams are emitted by
leds and sensed by photo-receivers
• light pulses are converted into
electrical pulses, electronically
Photo-receiver processed and transmitted in output
multiple
bit “race” o
Nt = # tracks = # bits zone • resolution = 360 / 2 Nt
Photo-receiver
(min 12 in robotics) • digital encoding of absolute position
when the optical disk is rotating fast, the
use of binary coding may lead to (large)
reading errors, in correspondence to
multiple transitions of bits

10
2
 Absolute encoding

DECIMAL BINARY GRAY

binary
coding
XOR
optical disks
with 2 bits

Gray
coding

GRAY BINARY
CODE CODE

8-bit Gray-coded adjacent codes differ

absolute encoder by just one bit
Robotics 1 10
3
 Use of absolute encoders
• ready to measure at start (no “homing”)
• two modes for permanent operation
- when switching off the drive, position
parameters are saved on a flash memory
(and brakes activated)
hollow shaft
- battery for the absolute encoder is always
active, and measures position even when
the drive is off
- data memory > 20 years
• single-turn or multi-turn versions, e.g. round flange
- 13-bit single-turn has 213 = 8192 steps per
revolution (resolution = 0.044o)
- 29-bit multi-turn has 8192 steps/revolution
+ counts up to 216 = 65536 revolutions
• aluminum case with possible interface to field
bus systems (e.g., CANopen or PROFIBUS)
13-bit absolute encoder opened: • typical supply 5/28V DC @1.2 W
Gray-coded disk and electronics multi-turn
10
4
 Incremental encoders
Photo-emitter
• optical rotating disk with three tracks, alternating
Optical disk transparent and opaque areas: measures
incremental angular displacements by counting trains
of Ne pulses (“counts”) per turn (Ne = 100÷5000)

• the two A and B tracks (channels) are

in quadrature (phase shift of 90o
electrical), allowing to detect the
direction of rotation
Photo-receiver
• a third track Z is used to define the
“0” reference position, with a reset of
the counter (needs “homing” at start)
The three tracks
on an optical disk
(here Ne = 6) • some encoders provide as output also
the three phases needed for the
switching circuit of brushless motors
10
5
 Incremental encoders

• two (cheap) incremental encoders • a OMRON incremental encoder

inside a mouse with 2000 pulses/turn

diameter ∅ 40 mm
mass m ≈ 100 g
inertia J = 1⋅10-6 kg m2

10
6
 Velocity sensor: Tachometer
always mounted on the (electrical) motor axis

principle of operation (single coil)


V
  
N S
B = cost
permanent magnet coil with area 
magnetic field:
flux through the coil is (B) = |B| cos   |B| cos t

V = - d/dt = |B|  sin t ⇒ to reduce ripples, use m coils

rotated regularly by 180o/m
amplitude V 
10
7
 Accelerometers
 measure of linear acceleration based on inertial forces (no “touch”)
 units: [m/s²] or gravitational acceleration [g] (non-SI unit: 1g ≈ 9.81 m/s²)
 different principles for converting mechanical motion in an electrical signal
 piezoelectric: piezoceramics (PZT) or crystals (quartz), better linearity & stability,
wide dynamic range up to high frequencies, no moving parts, no power needed
 piezoresistive: for high-shocks, measures also static acceleration (g0), needs supply
 capacitive: silicon micro-machined sensing element, superior in static to low
frequency range, can be operated in servo mode, cheap but limited resolution
 modern solution: small MEMS (Micro Electro-Mechanical Systems)
 multiple applications: from vibration analysis to long range navigation

animation of
measurement principle
in a piezoelectric
accelerometer

10
8
 Operation principle
seismic accelerometer
accelerometer spring
case

seismic piezoelectric
mass element
𝑑:𝑧 sensor measure
𝑎=
𝑑𝑡 : 𝑎𝑚 output
seismic mass
B
base © Brüel & Kjaer
surface of moving accelerometer
(or vibrating) body plate

𝐴 𝑚 (𝑠) 𝑀
= 𝐾𝑡
𝑀𝑎 = 𝑀B̈ + 𝐷Ḃ + 𝐾B by Laplace
𝐴(𝑠) 𝑀 𝑠 : + 𝐷𝑠 + 𝐾
𝑎𝑚 = 𝐾𝑡 B transform 𝐾M
= O
𝑠 P 𝐷⁄𝑀 𝑠P 𝐾⁄𝑀
10
9
 Frequency characteristics
of a piezoelectric accelerometer
U (𝑠) 𝐾M 𝐷
𝐹 𝑠 = 𝑚 = 𝜔𝑟 = 𝐾Y Z= 1Y
U(𝑠) 𝑠 O P 𝐷⁄𝑀 𝑠P 𝐾⁄𝑀 𝑀 2 𝐾𝑀
resonance damping
frequency factor

𝐹([𝜔) model valid

above this
frequency

1.1
0 dB = 1

- 3 db = 0.7071

low Bu r B-3
low frequency limit useful bandwidth 𝐵𝑢 ≈ 0.3𝜔𝑟 bandwidth
of piezoelectric use (≤ 1 Hz) for 10% error
11
0
 MEMS accelerometers
 very simple MEMS (a cantilever beam with a test mass, with damping from the
residual gas sealed in the device), single- or tri-axial, very small and light
 cross-couplings among acceleration sensing directions should be limited ≤ 3%

ADXL335 3-axis, small,

low power, ±3g, with signal
conditioned voltage outputs

11
1
Mounting accelerometers on robots

Bosch BMA 150 3-axial accelerometers

integrated in two larger Tactile Modules on the
links of a Bioloid humanoid left arm [TUM, 2011]

3-axial MEMS
accelerometer
on the forearm
of a KUKA KR15/2 3-axial capacitive accelerometer
[DLR/Sapienza, 2007] on end-effector tool of an ABB robot
(Crossbow Technology: 2g range,
1V/g output, 0-50 Hz, ± 2 ˚ align error)
[Linköping, 2012]
11
2
 Robot
components(Sensors):
Exteroceptive sensors

11
3
 External sensors
 force sensors
 strain gauges and joint torque sensor
 6D force/torque (F/T) sensor at robot wrist
 proximity/distance sensors
 infrared (IF)
 ultrasound (US)
 laser
 vision

Robotics 1 11
4
 Force/torque and deformation

 indirect information obtained from

the measure of deformation of an
elastic element subject to the force
or torque to be measured
 basic component is a strain gauge:
it uses the variation of the resistance
𝑅 of a metal conductor when its
length 𝐿 and/or cross-section 𝑆 vary

%𝑅 %𝑅 %𝑅 small
>0 <0
%𝐿 %𝑆 %𝑇
temperature
11
5
 Strain gauges
10 mm Wheatstone single-point quarter-bridge
(for accurately measuring resistance)

+
𝑉+ 𝑉𝑜 +

principal measurement axis

∆𝑅⁄𝑅 strain )
Gauge-Factor = GF = • 𝑅1, 𝑅2, 𝑅 ( very well matched (≈ 𝑅)
∆𝐿⁄𝐿
• 𝑅𝑆 ≈ 𝑅 at rest (no stress)
(typically GF ≈ 2, i.e., small sensitivity) • two-point bridges have 2 strain gauges
connected oppositely (➚ sensitivity)

if 𝑅1 has the same dependence on 𝑇 of 𝑅𝑆 𝑅( 𝑅2

thermal variations are automatically 𝑉=
1 − 𝑉+
compensated
𝑅 ( + 𝑅𝑠 𝑅1 + 𝑅2
11
6
 Strain gauges in flexible arms

7 strain gauges glued(1) to a flexible aluminum beam (a robot “link”)

measuring its local “curvature” in dynamic bending during slew motions
(a proprioceptive use of these sensors)

(1) by cyanoacrylic glue

11
7
 Torque sensor at robot joints
use of flexible bellow couplings
for aligning motors and long shafts

strain gauge mounted to “sense” the axial deformation

of the transmission shaft of joint #3 (elbow) in a PUMA 500 robot
(again, a proprioceptive use of this sensor)

11
8
 Force/torque sensor at robot wrist

 a device (with the outer form of a cylinder), typically located

between the last robot link and its end-effector.
 top and bottom plates are mechanically connected by a
number of deformable elements subject to strain under the
action of forces and moments
 there should be at least one such element in any direction
along/around which a force or torque measure is needed
 since a complete “decoupling” of these measurements is
hard to obtain, there are 𝑁 ≥ 6 such deformable elements
 on each element, a pair of strain gauges is glued so as to
undergo opposite deformations (e.g., traction/compression)
along the main axis of measurement

11
9
 Maltese-cross configuration

• diameter  10 cm • 4 deformable elements

• height  5 cm • two pairs of strain gauges are mounted on
• 50÷500 N (resolution 0.1%) opposite sides of each element (8 pairs)
• 5÷70 Nm (resolution 0.05%) • the two gauges of each pair are placed
• sample frequency  1 KHz adjacent on the same Wheatstone bridge

12
0
 6D force/torque sensors

 ATI series
 cost (in 2016): about 6 K€ for
Mini45 model + 700 € DAQ card

12
1
 6D force/torque sensor

 electronic processing unit and mounting on an industrial

robot (Comau Smart 3 robot, 6R kinematics)

mounting flange
(on link 6 of the manipulator arm)

12
2
 6D F/T sensor calibration

force/torque measured in the

frame attached to the sensor
calibration matrix
𝑐A
𝑐
𝑐𝑅
𝑠 𝑂 𝑠A
𝑠
output of
𝑐𝜇 = 𝑐𝑅 𝑠𝜇 Wheatstone
𝑐 𝑆 𝑐𝑟𝑐𝑠 𝑐𝑅𝑠 𝑠 𝑠
bridges
transformation from the sensor frame
to the load/contact frame (at TCP)
12
3
Typical uses of a F/T sensor

12
4
 Proximity/distance sensors - 4
 structured light: a laser beam (coherent light source) is projected on
the environment, and its planar intersection with surrounding objects
is detected by a (tilted) camera
 the position of the “red pixels” on the camera image plane is in
trigonometric relation with the object distance from the sensor

CCD camera image

object
laser beamer

side view top view

projected laser beams

(2D in this case)
32
 Use of structured light sensors

Random bin picking of 10-30 parts/minute (with

surface inspection) with a 6R industrial robot,
two pairs of cameras and a structured light sensor
[Universal Robotics]
Structured light approach to best fit and
finish car bodies (down to 0.1 mm)
for reducing wind noise
[Ford Motor Co.]

Virtobot system for post-mortem Hercules ROV + structured-laser-light

3D optical scanning of human body imaging system for high-resolution
& image-guided needle placement bathymetric underwater maps
[Univ. Zürich] [Univ. Rhode Island]
34
 Eye-in-hand camera

camera frame

robot manipulator

base frame

object frame

Relevant reference frames for visual-based tasks

42
 Kinect
camera + structured light 3D sensor
 RGB camera (with 640×480 pixel)
 depth sensor (by PrimeSense)
 infrared laser emitter
 infrared camera (with 320×240 pixel)
 30 fps data rate
 range: 0.5÷5 m
Kinect 2.0  depth resolution: 1cm@2m; 7cm@5m
(2014)  cost: < 90 €

“skeleton” extraction and

human motion tracking
46
 Kinect
Depth sensor operation

 stereo triangulation based on IR source

emitting pseudo-random patterns
 reference pattern on IR camera image
depths
plane acquired in advance from a plane
at known distance and coded in H/W
focal  correlating the disparity 𝑑 (10 bits) of
length reference and received object patterns
image plane baseline provides the object depth 𝑧𝑘
1. triangulation equations (by similarity of triangles)
𝑧a
𝐷 𝑧8 − 𝑧 a 𝑑 𝐷 𝑧8 𝑥a = −
A a
(N − N8 + _N)
= & = 𝑧a = 𝑧a
𝑏 𝑧8 A 𝑧a 1 + 𝑑 𝑧8 𝑦a = − (𝑌a − 𝑌8 + _𝑌)
A
A𝑏
2. accurate calibration of sensor
baseline length 𝑏, depth of reference 𝑧0 + camera intrinsic parameters
(focal length A, lens distortion coefficients _N, _𝑌 , center offsets N8, 𝑌0)
47
 Manipulators and vision systems

 stereovision with two external cameras, fixed in the

environment (eye-to-hand)

51
 Manipulators and vision systems

 CCD camera mounted on the robot for controlling the

end-effector positioning (eye-in-hand)

52
 Visual tracking eye-to-hand

COMAU robot with position-based 6D tracking from external camera

(DIS, Università di Napoli Federico II)
53
 Visual servoing eye-in-hand

video

Image-based servoing with camera mounted on the robot end-effector

(IRISA/INRIA, Rennes)
54
 Visual servoing and redundancy

visual servoing of circle feature (𝑚 = 3: 𝑝𝑥, 𝑝𝑦, 𝑟) by Adept Viper robot (𝑛 = 6):
redundancy is used for avoiding joint range limits (IRISA/INRIA, Rennes)
55
Combined visual/force assembly

KUKA LWR with eye-in-hand camera and F/T sensor

(DLR, IEEE ICRA’07 demo in Roma)
57