1

Robotics
Teacher:
Dr. Syed Riaz un Nabi Jafri

2
Moving Robots
• A robot having capability to change its
location is known as moving robot
• Variety of moving robots are available
Moving Robots
Structure
• Different kind of structures are in use
according to demand
• This presentation is covering Differential Drive
Rover which are most common structure for
indoor/outdoor environments
Structure

Main body
Joints

Controller

Actuators
Processor

Sensors
Moving Robot Components
• Rover: main body of robot
• Tx/Rx: Communication modules
• Joints: Provide connectivity between different parts
• Actuators: Agents which are responsible to provide force for motion on
different parts
• Sensors: Agents which are responsible to sense rate of motion of robot
itself or objects nearer to robot with further details
• End effector: The part at the last joint (hand/gripper)
• Controller: A numerically supported agent responsible to control all
motions of robot
• Processor (Brain): A numerically supported agent responsible to
determine and estimate behavior of robot and its surroundings and to
generate commands to different parts
• Software: A series of instructions which are provided to Processor
• Base station : For data logging and control
7
Block Diagram

Base Station

Motion
Sensors

Processor Controller Actuators

Perception
Sensors

8
Workspace

Could you please draw workspace ?

9
Movement Principle
Two Actuators are connected on each
wheel. A balancing wheel can be placed at
front or back side.

Rotation rate of each wheel can be

controlled through variable PWM signals.

Based on respective PWM values, robot can

move in forward, right or left directions.
Actuators

Differential drive robot

10
Movement Principle
One Actuator is connected on front shaft for
two wheels. It is responsible to turn right or
left the vehicle.

One Actuator is connected on back shaft for

two wheels. It is responsible to move
forward or backward the vehicle.

Rotation rate of each actuator can be

controlled through variable PWM signals. Actuators

Car like robot (Ackerman steering)

11
Examples

Car like robot (Ackerman steering)

Differential drive robot

12
Applications
• In use for indoor/outdoor cleaning
• For surveillance
• As an industrial loader
• For inspection
• Other domestic use

13
Degree Of Freedom (DOF)

Could you please let me know DOF ?

14
Actuators
Actuators are referred as Muscles of robot

• Electric motors
Servo motors
Stepper motors
• Pneumatic actuators
• Hydraulic actuators

15
Actuators

Pneumatic actuator
DC Servomotor

Hydraulic actuator
Stepper motor

16
Actuator Selection
• Select best actuator for following:

17
Sensors
• Potentiometers
• Encoders
• LVDT
• Resolvers
• GPS (Global positioning system)
• IMU (inertial measurement unit)
• Proximity Sensors
• Range Finders
• Object Identifiers

18
Sensors

Potentiometer
Encoder IMU

Resolver

GPS
19
Sensors

2D Laser range finder

2D ultrasonic range finder

20
Sensors

IR Sensor
Camera

Kinect (RGB-D)

21
Sensors Combination
• Position sensing (Localization)
Indoor ? (encoder, Pot, IMU,GPS)
Outdoor ?

• Environment perception (Mapping)

Indoor ? (encoder, Laser, IMU,Camera)
Outdoor ?

22
Processors (Brain)
A variety of processors solutions are available
depending on the requirement.

• Microprocessor based main boards

• Microcontroller based main boards
• FPGA based main boards

23
Processors (Brain)

ARM
Raspberry PI

FPGA

Tern
24
Reference Frame

25
Sensor Working Principle

XS

YS

2D laser sensor consists of laser LED and

XS photodiode/transistor. LED emits light and
diode detects its reflection from some
YS
object and determine distance using S =26V*T
Reference Frame

27
Objects points in Sensor Frame

Number of scan points could be more

depending on the objects
28
Objects features in Sensor Frame

29
Objects features in Robotic Frame

30
Objects features in World Frame
(Transformations)

31
Real Environment Features

Most of the scan points have merged into line segments, there
are some points which have not merged. This is the demerit of
feature based approach.
32
Scan points conversion into features
(Split & Merge)

Features handling is easy as compare to raw scan points

33
Transformations

Yw
Translation (dZ = 0)

Xw

Rotation along z-axis

34
 Transformations
Consider following wall W1 has
been sensed by the laser scanner
Example:
Robot center is placed at (5,5) W1
with orientation 40 degrees.
Scanner center has displacement
of 10 units from robot center.
Both Scanner and robot frames
5,5 10
are aligned with each other.

35
Transformations
Scanning values of the wall has been given
(4,6)

W1

(7,2)

XS

36
Transformations

We need to do following steps:

W1

• Translation (from sensor to robot frame)

• Rotation (robot frame to world frame)
• Translation (robot frame to world frame)

Pxyz = Trans(5,5,0) x Rot(Z,40 degrees) x Trans(10,0,0) x Ps

37
Transformations
Matlab script:
Ps => P_s = [4; 6; 0; 1]
Trans(10,0,0) => T2 = [1 0 0 10;0 1 0 0;0 0 1 0;0 0 0 1 ]

Rot(Z,40 degrees) => Rot = [cos(40*pi/180) -sin(40*pi/180) 0 0;

sin(40*pi/180) cos(40*pi/180) 0 0;
0 0 1 0;
0 0 0 1]

Trans(5,5,0) => T1 = [1 0 0 5;0 1 0 5;0 0 1 0;0 0 0 1]

Pxyz => P_xyz

P_xyz = T1 * Rot * T2 * P_s

38
 Transformations

Exercise:
Robot center is placed at (5,0) with orientation 0 degree. Sensor center
has displacement of 5 units from robot center. Both Sensor and robot
frames are aligned with each other. Robot sensor is detecting a line
end point at (5,0) in sensor frame. Determine end point coordinates in
world frame. Also verify result by plotting all frames manually.

39
Transformations

Exercise:
1
3 (1,1)

XR XS
3

XW

40
Transformations
Line parameters conversion
(rg = 4, ϕg = 90 degrees)
line

XR

6 , 0 degrees

XW

41
Ultrasonic Sensor Scans

Original scan Processed scan

42
Environment Perception (Features
based Map)

Features derived from raw scan points

43
Environment Perception (Grid based
Map)

Grids showing three states: occupied, non occupied,

undetermined
44
Localization
Finding location and orientation of robot within known
environment.

Localization could be local when we are comparing two

consecutive scans, in other case, it could be global when
we are comparing current scan with global map.

45
Localization
We compare two consecutive scans to
determine robot new pose

First Scan Second Scan (in blue)

World frame view

46
Localization
We compare current scan with given known
map to determine robot new pose

47
Simultaneous Localization and Mapping
(SLAM)

What is SLAM?
Estimate the pose of a robot and the map of the environment
at the same time
SLAM is hard, because
a map is needed for localization and
a good pose estimate is needed for mapping
Localization: inferring location given a map
Mapping: inferring a map given locations
SLAM: learning a map and locating the robot simultaneously
Various algos are available like EKF and particle filter SLAM

48
Simultaneous Localization and Mapping
(SLAM)

49
Simultaneous Localization and Mapping
(SLAM)

50
Simultaneous Localization and Mapping
(SLAM)

51
Simultaneous Localization and Mapping
(SLAM)

52
Simultaneous Localization and Mapping
(SLAM)

53
Environment Perception (3D)
Single Camera as Range & Bearing Sensor

54
Environment Perception (3D)

55
Environment Perception (3D)

56
Useful links
• Books:
Introduction to Robotics: Analysis, Control, Applications (By Saeed B. Niku)
Robotic Vision and Control (By Peter Corke)
Probabilistic Robotics (By Thrun et al)

• Journals/Conferences
IEEE Transactions on Robotics, Advanced Robotics, Journal of Field Robotics, IEEE ICRA, IEEE
IROS, IEEE ROBIO

• Online courses
MIT, Stanford, Coursera, EDX, IIT India

• Platforms
ROS , Webots, Vrep, OpenCV

• Groups
Robotics worldwide (search more specified)
57