AIE455 ROBOT

MAPPING AND
LOCALIZATION
DR.SAFA A.ELASKARY

FALL 2024-2025
SLAM Lecture Outline
01 • SYSTEM NAVIGATION REVIEW

02 • SLAM DEFINITION

03 SLAM CHALLENGES

SLAM SLAM
04 APPLICATIONS
05 SIMULATION

HOW SLAM
06 WORKS?
 2
01- SYSTEM NAVIGATION REVIEW
 01- SYSTEM NAVIGATION REVIEW
3
OUTLINES

-Robot Navigation (Planning Based Control)

-Perception
-Localization
-Cognition
-Motion Control
-Simultaneous Localization and Mapping Definition
 01- SYSTEM NAVIGATION REVIEW

Figure 1 – Planning Based Control

-Through perception, a model of the “real” world is captured

in memory.
-A goal is given and a plan is generated, assuming the “real”
world is not changing.
-Then, the plan is executed, one (abstract) operation at a
time.
 01- SYSTEM NAVIGATION REVIEW 5

Perception
Proprioceptive
-Compass
-Encoders
-Accelerometers
-IMU – Inertial Measurement Unit

Figure 2 – Proprioceptive Sensors

 6
01- SYSTEM NAVIGATION REVIEW

Perception
Exteroceptive
-Vision Systems (LIDAR)
-Range Sensors
-Positioning System (e.g. GPS)

Figure 3 – Exteroceptive Sensors

 Fall 2024-2025

8 LIDAR (LIGHT DETECTION AND RANGING)

01- SYSTEM NAVIGATION REVIEW 7

Localization
Representation
-Continuous
-Discrete
Figure 4 – Representation

Probabilistic
Representations
-Continuous
-Discrete
Figure 5 – Probabilistic Representation
01- SYSTEM NAVIGATION REVIEW 8

Figure 6– Kalman Filter Algorithm

 02- SLAM DEFINITION
 SLAM stands for simultaneous localization and mapping

 The task of building a map while estimating the pose

of the robot relative to this map
 The backbone of spatial awareness of a robot
 Traditional SLAM: Pick natural scene features as
landmarks, observe their motion & reason about robot
motion

Why is SLAM hard?

 Chicken and egg problem:

a map is needed to localize the robot
and a pose estimate is needed
to build a map
11 Fall 2024-2025
 02- SLAM DEFINITION

• Robot localization and mapping: are fundamental capabilities for mobile robots
operating in there all world. Even more challenging than these individual problems is
their combination: simultaneous localization and mapping(SLAM) .

• An accurate pose estimate is necessary for building a map of the environment

Mapping with known robot poses.

• We will explore suitable efficient probabilistic inference algorithms at the

intersection of linear algebra and probabilistic graphical models

12
Fall 2024-2025
 Fall 2024-2025

02- SLAM DEFINITION

WHEN IS IT NECESSARY?

• When a robot must be truly autonomous(no human input).

• When there is no prior knowledge about the environment.

• When we cannot rely external positioning systems (e.g. GPS).
• When the robot needs to know where it is.

13
 03- SLAM CHALLENGES
A robot moving through an unknown, static environment

Given:
• The robot’s controls
• Observations of nearby features

Estimate:
• Map of features
• Path of the robot

14
 03- SLAM CHALLENGES
WHY IS SLAM A HARD PROBLEM?

SLAM: robot path and map are both unknown!

Robot path error correlates errors in the map
Fall 2024-2025
15
 Fall 2024-2025

16
03- SLAM CHALLENGES

Robot pose
uncertainty

• In the real world, the mapping between observations and

landmarks is unknown
• Picking wrong data associations can
catastrophic consequences
• Pose error correlates data associations/*
 17 DATA ASSOCIATION PROBLEM

• A data association is an assignment of

observations to landmarks
• In general there are more than [n choose m]
(n observations, m landmarks) possible
associations
• Also called “assignment problem”

Fall 2024-2025
Implementing Simultaneous Localization and Mapping (SLAM) for
autonomous vehicles presents several challenges that can impact the
accuracy and reliability of navigation systems. Here are the main challenges
identified in the search results:

1. Noisy Sensor Data

2. Computational Complexity
3. Accumulation of Localization Errors
4. Dynamic Environments
5. Calibration of Multiple Sensors
6. Environmental Limitations

18

Fall 2024-2025
 Fall 2024-2025

19 1.NOISY SENSOR DATA

• SLAM relies on data from various sensors, such as LiDAR,

cameras, and IMUs (Inertial Measurement Units). However,
these sensors often produce noisy data due to
environmental factors or inherent sensor limitations. The
noise can lead to inaccuracies in both localization and
mapping, making it difficult for the vehicle to maintain an
accurate understanding of its surroundings
 Fall 2024-2025

20 2.COMPUTATIONAL COMPLEXITY

• The computational cost associated with processing sensor

data is significant. SLAM algorithms require high-frequency
image processing, point cloud matching, and optimization
calculations, which can be demanding for the limited
processing power available in vehicle hardware. This high
computational demand can hinder real-time performance,
especially in dynamic environments where quick decision-
making is crucial
Fall 2024-2025

3. ACCUMULATION OF LOCALIZATION
ERRORS

• As the vehicle moves, small errors in localization can

accumulate over time, leading to substantial deviations from
the actual position.
• This issue is particularly pronounced in environments
where loop closure is necessary to correct these errors.
• If not managed properly, accumulated errors can distort the
map and make it challenging for the vehicle to navigate
accurately

21
Fall 2024-2025

4. DYNAMIC ENVIRONMENTS

• Autonomous vehicles often operate in highly dynamic

environments with moving objects (e.g., pedestrians, other
vehicles).
• This variability complicates the SLAM process as the
algorithms must continuously adapt to changes while
maintaining accurate localization and mapping.
• The presence of dynamic obstacles can lead to
misinterpretation of the environment and affect the
vehicle's ability to navigate safely

22
 Fall 2024-2025

23 5. CALIBRATION OF MULTIPLE
SENSORS

• Using multiple sensors for SLAM can enhance robustness

but also introduces challenges related to calibration.
• Discrepancies between sensor readings can lead to
inaccuracies in data fusion, undermining the overall
effectiveness of the SLAM system.
• Accurate calibration is essential to ensure that all sensors
work harmoniously
Fall 2024-2025

6. ENVIRONMENTAL LIMITATIONS

• Certain environmental conditions—such as poor lighting or

lack of distinct features—can hinder the effectiveness of
visual SLAM systems.
• For instance, LiDAR systems may struggle when there are
few obstacles or when obstacles are too far away, leading to
difficulties in aligning point clouds and maintaining accurate
localization

24
 04 - SLAM APPLICATIONS

• SLAM algorithms are useful in many other applications

such as navigating a fleet of mobile robots to arrange
shelves in a warehouse, parking a self-driving car in an
empty spot, or delivering a package by navigating a
drone in an unknown environment.

25
Fall 2024-2025
04 - SLAM APPLICATIONS

Indoors Undersea

Space Underground

Fall 2024-2025
26
 27
Fall 2024-2025
Consider a home robot vacuum. Without SLAM, it will just move randomly within a room
and maybe unable to clean the entire floor surface. In addition, this approach uses excessive
power, so the battery will run out more quickly. On the other hand, robots with a SLAM
algorithm can use information such as the number of wheel revolutions and data from cameras
and other imaging sensors to determine the amount of movement needed. This is called
localization. The robot can simultaneously use the camera and other sensors to map the
obstacles in its surroundings and avoid cleaning the same area twice. This is called mapping.

28
Fall 2024-2025
Fall 2024-2025
29
 30
Fall 2024-2025
 31
Fall 2024-2025
 Fall 2024-2025

32 05 - SLAM SIMULATION

• MATLAB® and Simulink® provide SLAM algorithms,

functions, and analysis tools to develop various
applications. You can implement simultaneous
localization and mapping along with other tasks
such as sensor fusion, object tracking path
planning, and path following.
 Fall 2024-2025

33 06 - HOW SLAM WORKS?

There are two types of technology components used to

achieve SLAM.
The first type is sensor signal processing, including the front-end
processing, which is largely dependent on the sensors used.
The second type is pose-graph optimization, including the back-end
processing, which is sensor-agnostic.
06 - HOW SLAM WORKS?

SLAM PROCESSING FLOW

34
Fall 2024-2025
BROADLY SPEAKING, THERE ARE TWO TYPES OF
TECHNOLOGY COMPONENTS USED TO ACHIEVE
SLAM.

• sensor signal processing,

The first • the front-end processing
type • back-end processing,

• pose-graph optimization
The second • back-end processing,
type • sensor-agnostic.

35
Fall 2024-2025
 Fall 2024-2025

36 CONCLUSION

• SLAM is a hard problem which is not yet fully solved

• These challenges highlight the complexity of
implementing SLAM in autonomous vehicles.
Ongoing research and development focus on
addressing these issues through improved
algorithms, better sensor technologies, and
enhanced computational methods to ensure safer
and more reliable autonomous navigation systems.
• Probabilistic methods which take account of sensor
and process model error tend to work best
 Fall 2024-2025

37
REFERENCES ON EKF SLAM
• P. Moutarlier, R. Chatila, "Stochastic Multisensory Data Fusion for Mobile
Robot Localization and Environment Modelling", In Proc. of the International
Symposium on Robotics Research, Tokyo, 1989.

• R. Smith, M. Self, P. Cheeseman, "Estimating Uncertain Spatial

Relationships in Robotics", In Autonomous Robot Vehicles, I. J. Cox and G.
T. Wilfong, editors, pp. 167-193, Springer-Verlag, 1990.

• Ali Azarbayejani, Alex P. Pentland, "Recursive Estimation of Motion,

Structure, and Focal Length," IEEE Transactions on Pattern Analysis and
Machine Intelligence ,vol. 17, no. 6, pp. 562-575, June, 1995.
 Fall 2024-2025

38
ADDITIONAL REFERENCE
• https://www.mathworks.com/discovery/slam.html

• http://www.probabilistic-robotics.org
Fall 2024-2025

39

• Thank you