ROBOTICS​ ​SOFTWARE​ ​ENGINEER

NANODEGREE​ ​SYLLABUS

Course​ ​1:​ ​Introduction​ ​to​ ​Robotics 2

Course​ ​2:​ ​ROS​ ​Essentials 2

Course​ ​3:​ ​Kinematics 3

Course​ ​4:​ ​Perception 3

Course​ ​5:​ ​Controls 4

Course​ ​6:​ ​Deep​ ​Learning 5

Course​ ​7:​ ​Term​ ​2​ ​Introduction 6

Course​ ​8:​ ​Robotic​ ​Systems​ ​Deployment 6

Course​ ​9:​ ​Localization 7

Course​ ​10:​ ​SLAM 8

Course​ ​11:​ ​Reinforcement​ ​Learning​ ​for​ ​Robotics 8

Course​ ​12:​ ​Path​ ​Planning​ ​and​ ​Navigation 9

Course​ ​1:​ ​Introduction​ ​to​ ​Robotics
In​ ​this​ ​course,​ ​you'll​ ​get​ ​an​ ​introduction​ ​to​ ​your​ ​Nanodegree​ ​program,​ ​and​ ​explore​ ​the​ ​three​ ​essential
elements​ ​of​ ​robotics:​ ​perception,​ ​decision​ ​making,​ ​and​ ​actuation.​ ​You'll​ ​also​ ​build​ ​your​ ​first​ ​project,​ ​which
is​ ​modeled​ ​after​ ​the​ ​NASA​ ​Mars​ ​Rover​ ​Challenge.

Lesson Title Description

1 Welcome In​ ​this​ ​first​ ​lesson,​ ​ ​you'll​ ​meet​ ​your​ ​instructors,​ ​learn​ ​about
the​ ​structure​ ​of​ ​this​ ​program​ ​and​ ​about​ ​the​ ​services
available​ ​to​ ​you​ ​as​ ​a​ ​student.

2 What​ ​is​ ​a​ ​Robot? Ask​ ​three​ ​people​ ​what​ ​a​ ​robot​ ​is​ ​and​ ​you'll​ ​get​ ​three
different​ ​answers!​ ​ ​Here​ ​we​ ​ask​ ​your​ ​instructors,​ ​three
expert​ ​roboticists​ ​from​ ​Electric​ ​Movement.

3 Search​ ​and​ ​Sample​ ​Return In​ ​this​ ​lesson,​ ​you'll​ ​learn​ ​the​ ​skills​ ​you​ ​need​ ​to​ ​tackle​ ​the
first​ ​project,​ ​where​ ​you'll​ ​experience​ ​the​ ​three​ ​essential
elements​ ​of​ ​robotics​ ​--​ ​perception,​ ​decision​ ​making,​ ​and
actuation.

Project​ ​1​ ​:​ ​Search​ ​and​ ​Sample​ ​Return

In​ ​this​ ​project,​ ​you​ ​will​ ​write​ ​code​ ​to​ ​autonomously​ ​map​ ​a​ ​simulated​ ​environment​ ​and​ ​search​ ​for
samples​ ​of​ ​interest.

4 Careers:​ ​Orientation As​ ​you​ ​learn​ ​the​ ​skills​ ​you’ll​ ​need​ ​in​ ​order​ ​to​ ​work​ ​in​ ​the
robotics​ ​industry,​ ​you’ll​ ​see​ ​optional​ ​Career​ ​Lessons​ ​and
Projects​ ​that​ ​will​ ​help​ ​you​ ​prepare​ ​for​ ​interviews,​ ​craft​ ​your
resumé​ ​and​ ​more.

Course​ ​2:​ ​ROS​ ​Essentials

ROS​ ​provides​ ​a​ ​flexible​ ​and​ ​unified​ ​software​ ​environment​ ​for​ ​developing​ ​robots​ ​in​ ​a​ ​modular​ ​and
reusable​ ​manner.​ ​In​ ​this​ ​course,​ ​you'll​ ​learn​ ​how​ ​to​ ​manage​ ​existing​ ​ROS​ ​packages​ ​within​ ​a​ ​project,​ ​and
how​ ​to​ ​write​ ​ROS​ ​Nodes​ ​of​ ​your​ ​own​ ​in​ ​Python.

Lesson Title Description

 1 Introduction​ ​to​ ​ ​ROS Obtain​ ​an​ ​architectural​ ​overview​ ​of​ ​the​ ​Robot​ ​Operating
System​ ​Framework​ ​and​ ​set​ ​up​ ​your​ ​own​ ​ROS​ ​environment
on​ ​your​ ​computer.

2 Packages​ ​&​ ​Catkin Learn​ ​about​ ​ROS​ ​workspace​ ​structure,​ ​essential​ ​command
Workspaces line​ ​utilities,​ ​and​ ​how​ ​to​ ​manage​ ​software​ ​packages​ ​within
a​ ​project.​ ​Harnessing​ ​these​ ​will​ ​be​ ​key​ ​to​ ​building
shippable​ ​software​ ​using​ ​ROS.

3 Write​ ​ROS​ ​Nodes ROS​ ​Nodes​ ​are​ ​a​ ​key​ ​abstraction​ ​that​ ​allows​ ​a​ ​robot
system​ ​to​ ​be​ ​built​ ​modularly.​ ​In​ ​this​ ​lesson,​ ​ ​you'll​ ​learn
how​ ​to​ ​write​ ​them​ ​using​ ​Python.

Course​ ​3:​ ​Kinematics

Movement​ ​is​ ​one​ ​of​ ​the​ ​most​ ​exciting​ ​elements​ ​of​ ​building​ ​a​ ​Robot​ ​that​ ​interacts​ ​with​ ​the​ ​physical​ ​world.
In​ ​this​ ​course,​ ​you'll​ ​learn​ ​to​ ​control​ ​a​ ​robotic​ ​arm​ ​with​ ​six​ ​degrees​ ​of​ ​freedom​ ​to​ ​perform​ ​pick​ ​and​ ​place
actions​ ​using​ ​Inverse​ ​Kinematics.

Lesson Title Description

1 Intro​ ​to​ ​Kinematics In​ ​this​ ​lesson​ ​you'll​ ​get​ ​an​ ​introduction​ ​to​ ​the​ ​exciting​ ​field
of​ ​kinematics​ ​and​ ​learn​ ​about​ ​the​ ​most​ ​important​ ​types​ ​of
serial​ ​manipulators​ ​used​ ​in​ ​the​ ​robotics​ ​industry.

2 Forward​ ​and​ ​Inverse Here​ ​you'll​ ​dive​ ​deep​ ​into​ ​the​ ​details​ ​of​ ​solving​ ​the​ ​forward
Kinematics and​ ​inverse​ ​kinematics​ ​problem​ ​for​ ​serial​ ​manipulators.

3 Robotic​ ​Arm:​ ​Pick​ ​&​ ​Place In​ ​this​ ​lesson​ ​you​ ​will​ ​learn​ ​how​ ​to​ ​control​ ​a​ ​robotic​ ​arm
with​ ​six​ ​degrees​ ​of​ ​freedom​ ​to​ ​perform​ ​pick​ ​and​ ​place
actions​ ​using​ ​Inverse​ ​Kinematics.

Project​ ​2​ ​:​ ​Robotic​ ​Arm:​ ​Pick​ ​&​ ​Place

In​ ​this​ ​project,​ ​you​ ​will​ ​write​ ​code​ ​to​ ​perform​ ​Inverse​ ​Kinematics.​ ​Given​ ​a​ ​list​ ​of​ ​end-effector​ ​poses,​ ​you
will​ ​calculate​ ​joint​ ​angles​ ​for​ ​the​ ​Kuka​ ​KR210.

Course​ ​4:​ ​Perception

Robots​ ​perceive​ ​the​ ​world​ ​around​ ​them​ ​by​ ​using​ ​sensors.​ ​Working​ ​with​ ​sensor​ ​data​ ​for​ ​perception​ ​is​ ​a
core​ ​element​ ​of​ ​robotics.​ ​Here​ ​you'll​ ​work​ ​with​ ​3D​ ​point​ ​cloud​ ​data​ ​to​ ​perform​ ​segmentation​ ​tasks​ ​using
techniques​ ​like​ ​RANSAC​ ​and​ ​clustering.

Lesson Title Description

1 Perception​ ​Overview Here's​ ​a​ ​quick​ ​look​ ​at​ ​what​ ​to​ ​expect​ ​in​ ​the​ ​upcoming
lessons​ ​and​ ​project.

2 Introduction​ ​to​ ​3D Dive​ ​into​ ​the​ ​world​ ​of​ ​perception​ ​in​ ​three​ ​dimensions!​ ​After
Perception a​ ​brief​ ​tour​ ​of​ ​3D​ ​sensors​ ​used​ ​in​ ​robotics​ ​we'll​ ​explore​ ​the
capabilities​ ​of​ ​RGB-D​ ​cameras,​ ​which​ ​you'll​ ​use​ ​in​ ​these
lessons.

3 Calibration,​ ​Filtering,​ ​and To​ ​understand​ ​your​ ​sensor​ ​data​ ​you​ ​first​ ​need​ ​to​ ​calibrate!
Segmentation Here​ ​you'll​ ​get​ ​a​ ​handle​ ​on​ ​RGB-D​ ​camera​ ​calibration​ ​and
how​ ​to​ ​do​ ​filtering​ ​and​ ​basic​ ​segmentation​ ​on​ ​your​ ​point
cloud​ ​data.

4 Clustering​ ​for Clustering​ ​is​ ​a​ ​powerful​ ​machine​ ​learning​ ​method​ ​for
Segmentation segmenting​ ​objects​ ​of​ ​any​ ​arbitrary​ ​shape​ ​in​ ​your​ ​point
cloud​ ​data.​ ​ ​Here​ ​you'll​ ​compare​ ​K-means​ ​and​ ​Euclidean
clustering​ ​for​ ​object​ ​segmentation.

5 Object​ ​Recognition In​ ​this​ ​lesson,​ ​you'll​ ​take​ ​your​ ​segmented​ ​point​ ​cloud​ ​and
isolate​ ​features​ ​you​ ​can​ ​use​ ​to​ ​train​ ​a​ ​machine​ ​learning
algorithm​ ​to​ ​recognize​ ​the​ ​object​ ​you're​ ​looking​ ​for!

6 3D​ ​Perception​ ​Project In​ ​the​ ​project​ ​at​ ​the​ ​end​ ​of​ ​this​ ​lesson,​ ​you'll​ ​bring​ ​together
everything​ ​you​ ​know​ ​about​ ​perception​ ​in​ ​three​ ​dimensions,
from​ ​filtering​ ​and​ ​segmentation​ ​to​ ​feature​ ​extraction​ ​and
object​ ​recognition!

​ ​Project​ ​3:​ ​3D​ ​Perception

In​ ​this​ ​project,​ ​you​ ​will​ ​complete​ ​a​ ​tabletop​ ​pick​ ​and​ ​place​ ​operation​ ​using​ ​PR2​ ​in​ ​simulation.​ ​The​ ​PR2
is​ ​a​ ​common​ ​hardware​ ​and​ ​software​ ​platform​ ​for​ ​robot​ ​researchers.​ ​This​ ​one​ ​has​ ​been​ ​outfitted​ ​with​ ​a
noisy​ ​RGB-D​ ​sensor​ ​that​ ​your​ ​robot​ ​must​ ​use​ ​to​ ​identify​ ​and​ ​acquire​ ​objects​ ​from​ ​a​ ​cluttered​ ​space.

Course​ ​5:​ ​Controls

Control​ ​systems​ ​are​ ​a​ ​central​ ​component​ ​to​ ​most​ ​robots.​ ​In​ ​this​ ​course,​ ​you’ll​ ​learn​ ​how​ ​a​ ​mechanical
system​ ​can​ ​be​ ​described​ ​in​ ​terms​ ​of​ ​the​ ​equations​ ​that​ ​govern​ ​it.​ ​You'll​ ​then​ ​learn​ ​how​ ​to​ ​manage​ ​the
behavior​ ​of​ ​the​ ​system​ ​using​ ​a​ ​controller.​ ​Lastly,​ ​you’ll​ ​have​ ​an​ ​opportunity​ ​to​ ​observe​ ​your​ ​controller​ ​in
simulation.

Lesson Title Description

1 Introduction​ ​to​ ​Controls In​ ​this​ ​lesson​ ​you'll​ ​learn​ ​about​ ​Controls​ ​and​ ​how​ ​to​ ​create
and​ ​tune​ ​PID​ ​controllers.

2 Quadrotor​ ​Control​ ​using In​ ​this​ ​lesson,​ ​you'll​ ​learn​ ​how​ ​to​ ​control​ ​a​ ​Quadrotor​ ​inside
PID a​ ​Unity​ ​environment​ ​using​ ​a​ ​PID​ ​based​ ​Positional
Controller​ ​within​ ​a​ ​ROS​ ​node.

Course​ ​6:​ ​Deep​ ​Learning

Many​ ​recent​ ​developments​ ​in​ ​robotics​ ​can​ ​be​ ​attributed​ ​to​ ​advances​ ​in​ ​Deep​ ​Learning.
In​ ​this​ ​course,​ ​you’ll​ ​learn​ ​about​ ​Convolutional​ ​Neural​ ​Networks​ ​(CNN),​ ​Fully​ ​Convolutional​ ​Networks
(FCN),​ ​and​ ​Semantic​ ​Segmentation.​ ​ ​You​ ​will​ ​then​ ​integrate​ ​a​ ​Deep​ ​Neural​ ​Network​ ​with​ ​a​ ​simulated
quadcopter​ ​drone.

Lesson Title Description

1 Intro​ ​to​ ​Neural​ ​Networks In​ ​this​ ​lesson,​ ​Luis​ ​Serrano​ ​provides​ ​you​ ​with​ ​a​ ​solid
foundation​ ​for​ ​understanding​ ​how​ ​you​ ​build​ ​powerful​ ​neural
networks​ ​from​ ​the​ ​ground​ ​up.

2 TensorFlow​ ​for​ ​Deep Vincent​ ​Vanhoucke,​ ​Principal​ ​Scientist​ ​at​ ​Google​ ​Brain,
Learning introduces​ ​you​ ​to​ ​deep​ ​learning​ ​and​ ​TensorFlow,​ ​Google's
deep​ ​learning​ ​framework.

3 Deep​ ​Neural​ ​Networks Vincent​ ​walks​ ​you​ ​through​ ​how​ ​to​ ​go​ ​from​ ​a​ ​simple​ ​neural
network​ ​to​ ​a​ ​deep​ ​neural​ ​network.​ ​You'll​ ​learn​ ​about​ ​why
additional​ ​layers​ ​can​ ​help​ ​and​ ​how​ ​to​ ​prevent​ ​overfitting.

4 Convolutional​ ​Neural Vincent​ ​explains​ ​the​ ​theory​ ​behind​ ​Convolutional​ ​Neural

Networks Networks​ ​and​ ​shows​ ​you​ ​how​ ​to​ ​dramatically​ ​improve
performance​ ​in​ ​image​ ​classification.

5 Fully​ ​Convolutional In​ ​this​ ​lesson,​ ​you'll​ ​learn​ ​the​ ​motivation​ ​for​ ​Fully
Networks Convolutional​ ​Networks​ ​and​ ​how​ ​they​ ​are​ ​structured.

6 Semantic​ ​Segmentation In​ ​this​ ​lesson​ ​you'll​ ​be​ ​introduced​ ​to​ ​the​ ​problem​ ​of​ ​Scene
Understanding​ ​and​ ​the​ ​role​ ​FCNs​ ​play.

7 Project:​ ​Follow​ ​Me How​ ​to​ ​setup​ ​your​ ​environment​ ​and​ ​collect​ ​data​ ​for​ ​the
 Follow​ ​Me​ ​project.

Project​ ​4:​ ​Follow​ ​Me

In​ ​this​ ​project,​ ​you​ ​will​ ​build​ ​and​ ​train​ ​a​ ​Fully​ ​Convolutional​ ​Network​ ​(FCN)​ ​to​ ​find​ ​a​ ​specific​ ​sim-person
in​ ​images​ ​using​ ​semantic​ ​segmentation.​ ​Your​ ​simulated​ ​quadcopter​ ​will​ ​then​ ​run​ ​your​ ​trained​ ​model
with​ ​an​ ​inference​ ​engine​ ​in​ ​real​ ​time,​ ​to​ ​find​ ​the​ ​sim-person​ ​in​ ​video​ ​as​ ​the​ ​quadcopter​ ​patrols,​ ​and
follows.

8 Term​ ​1​ ​Outro Wrapping​ ​up​ ​your​ ​first​ ​term!

Course​ ​7:​ ​Introduction​ ​to​ ​Term​ ​2

In​ ​this​ ​course,​ ​you’ll​ ​get​ ​an​ ​introduction​ ​to​ ​Term​ ​2,​ ​and​ ​explore​ ​hardware​ ​commonly​ ​used​ ​in​ ​robotics.
You’ll​ ​learn​ ​the​ ​uses​ ​of​ ​common​ ​sensors,​ ​and​ ​which​ ​ROS​ ​packages​ ​you​ ​need​ ​to​ ​support​ ​them.​ ​ ​Those
students​ ​with​ ​TX2​ ​hardware​ ​will​ ​learn​ ​how​ ​to​ ​set​ ​up​ ​the​ ​system​ ​and​ ​interact​ ​with​ ​external​ ​hardware.

Lesson Title Description

1 Introduction​ ​to​ ​Term​ ​2 Term​ ​2​ ​brings​ ​a​ ​new​ ​direction​ ​to​ ​the​ ​program.​ ​ ​Robotics
applications​ ​today​ ​are​ ​adding​ ​machine​ ​learning​ ​techniques
to​ ​the​ ​traditional​ ​robotics​ ​portfolio.​ ​ ​You’ll​ ​learn​ ​the​ ​latest
reinforcement​ ​learning​ ​techniques​ ​as​ ​well​ ​as​ ​how​ ​they​ ​can
be​ ​deployed​ ​on​ ​TX2​ ​hardware​ ​platforms.

2 Introduction​ ​to​ ​the​ ​TX2 Students​ ​will​ ​also​ ​get​ ​a​ ​brief​ ​introduction​ ​to​ ​the​ ​Jetson​ ​TX2
and​ ​how​ ​to​ ​ ​setup​ ​the​ ​system.

3 Interacting​ ​with​ ​Robotics Introduce​ ​students​ ​to​ ​simple​ ​hardware​ ​I/O​ ​connections,
Hardware communication,​ ​and​ ​simple​ ​electrical​ ​theory.

4 Lab:​ ​Hardware​ ​Hello​ ​World Everyone​ ​who​ ​learns​ ​programming​ ​starts​ ​with​ ​a​ ​basic​ ​Hello
World​ ​program.​ ​In​ ​hardware,​ ​the​ ​Hello​ ​World​ ​version​ ​is​ ​to
blink​ ​an​ ​LED.​ ​Learn​ ​how​ ​to​ ​do​ ​that​ ​with​ ​the​ ​TX2.

5 Introduction​ ​to​ ​Robotics Brief​ ​introduction​ ​to​ ​various​ ​common​ ​robotics​ ​sensors,​ ​their
Sensor​ ​Options I/O,​ ​and​ ​their​ ​purpose.
Course​ ​8:​ ​Robotic​ ​Systems​ ​Deployment
In​ ​this​ ​course,​ ​you’ll​ ​learn​ ​new​ ​tools,​ ​and​ ​the​ ​embedded​ ​workflow​ ​as​ ​you​ ​move​ ​from​ ​code​ ​on​ ​a​ ​host
system​ ​to​ ​code​ ​on​ ​a​ ​target​ ​system.​ ​ ​You’ll​ ​work​ ​through​ ​a​ ​familiar​ ​problem​ ​with​ ​these​ ​new​ ​tools,​ ​then
extend​ ​what​ ​you’ve​ ​learned​ ​in​ ​a​ ​project.

Lesson Title Description

1 TX2​ ​Development Meet​ ​Kelvin​ ​Lwin​ ​of​ ​Nvidia.​ ​ ​Following​ ​Nvidia’s​ ​ ​“Two​ ​Days
to​ ​a​ ​Demo”,​ ​tutorial,​ ​learn​ ​about​ ​the​ ​tools​ ​and​ ​workflow​ ​for
developing​ ​and​ ​deploying​ ​robotics​ ​software​ ​to​ ​the​ ​TX2.

2 Inference​ ​Applications​ ​in Learn​ ​about​ ​the​ ​many​ ​and​ ​varied​ ​applications​ ​for​ ​inference
Robotics engines​ ​in​ ​robotics​ ​and​ ​the​ ​real-time​ ​considerations​ ​for
these​ ​systems.

3 Project:​ ​Robotic​ ​Inference Learn​ ​the​ ​steps​ ​to​ ​set​ ​up​ ​data​ ​and​ ​tune​ ​a​ ​deep​ ​neural
network​ ​with​ ​DIGITS.​ ​ ​Then​ ​apply​ ​this​ ​to​ ​your​ ​own​ ​version
of​ ​the​ ​Robotic​ ​Inference​ ​project.

Project​ ​5:​ ​Robotic​ ​Inference

Design​ ​your​ ​own​ ​robotic​ ​system​ ​using​ ​inference.​ ​You​ ​will​ ​create​ ​a​ ​project​ ​idea,​ ​collect​ ​your​ ​own​ ​data
set​ ​for​ ​classification,​ ​and​ ​justify​ ​network​ ​design​ ​choices​ ​based​ ​on​ ​your​ ​analysis​ ​of​ ​accuracy​ ​and​ ​speed
on​ ​the​ ​target​ ​system.

Course​ ​9:​ ​Localization

Learn​ ​how​ ​Gaussian​ ​filters​ ​can​ ​be​ ​used​ ​to​ ​estimate​ ​noisy​ ​sensor​ ​readings,​ ​and​ ​how​ ​to​ ​estimate​ ​a​ ​robot’s
pose​ ​relative​ ​to​ ​a​ ​known​ ​map​ ​of​ ​the​ ​environment​ ​with​ ​Monte​ ​Carlo​ ​Localization​ ​(MCL).

Lesson Title Description

1 Introduction​ ​to​ ​Localization Learn​ ​what​ ​it​ ​means​ ​to​ ​localize​ ​and​ ​the​ ​challenges​ ​behind
it.

2 Kalman​ ​Filters Learn​ ​what​ ​a​ ​Kalman​ ​filter​ ​is,​ ​and​ ​its​ ​importance​ ​in
estimating​ ​noisy​ ​data.

3 Lab:​ ​Kalman​ ​Filters Implement​ ​an​ ​Extended​ ​Kalman​ ​Filter​ ​package​ ​with​ ​ROS​ ​to
estimate​ ​the​ ​position​ ​of​ ​a​ ​robot.

4 Monte​ ​Carlo​ ​Localization Introduction​ ​to​ ​the​ ​MCL​ ​(Monte​ ​Carlo​ ​Localization)
algorithm​ ​to​ ​localize​ ​robots.
 5 Build​ ​MCL​ ​in​ ​C++ Learn​ ​how​ ​to​ ​code​ ​the​ ​MCL​ ​algorithm​ ​in​ ​C++

6 Project:​ ​Where​ ​am​ ​I? Set​ ​up​ ​and​ ​explore​ ​the​ ​steps​ ​for​ ​the​ ​Where​ ​am​ ​I?​ ​Project
using​ ​MCL​ ​with​ ​ROS​ ​in​ ​C++.

Project​ ​6:​ ​Where​ ​am​ ​I?

You​ ​will​ ​use​ ​the​ ​Monte​ ​Carlo​ ​Localization​ ​algorithm​ ​in​ ​ROS​ ​in​ ​conjunction​ ​with​ ​sensor​ ​data​ ​and​ ​a​ ​map
of​ ​the​ ​world​ ​to​ ​estimate​ ​a​ ​mobile​ ​robot’s​ ​position​ ​and​ ​orientation​ ​so​ ​that​ ​your​ ​robot​ ​can​ ​answer​ ​the
question​ ​“Where​ ​am​ ​I?”

Course​ ​10:​ ​ ​SLAM

Learn​ ​how​ ​to​ ​create​ ​a​ ​Simultaneous​ ​Localization​ ​and​ ​Mapping​ ​(SLAM)​ ​implementation​ ​with​ ​ROS
packages​ ​and​ ​C++.​ ​You’ll​ ​achieve​ ​this​ ​by​ ​combining​ ​mapping​ ​algorithms​ ​with​ ​what​ ​you​ ​learned​ ​in​ ​the
localization​ ​lessons.

Lesson Title Description

1 Mapping​ ​Algorithms Learn​ ​about​ ​probabilistic​ ​occupancy​ ​grid​ ​mapping.

2 Combining​ ​Localization​ ​and The​ ​intuition​ ​and​ ​conceptual​ ​background​ ​of​ ​Simultaneous
Mapping​ ​for​ ​SLAM Localization​ ​and​ ​Mapping​ ​(SLAM).

3 SLAM​ ​and​ ​ROS Learn​ ​about​ ​SLAM​ ​packages​ ​available​ ​in​ ​ROS​ ​and​ ​how​ ​to
use​ ​them.

4 Project:​ ​Map​ ​My​ ​World Set​ ​up​ ​and​ ​explore​ ​the​ ​steps​ ​you​ ​need​ ​to​ ​do​ ​the​ ​project
Robot with​ ​SLAM​ ​and​ ​ROS​ ​in​ ​C++.

Project​ ​7:​ ​Map​ ​My​ ​World​ ​Robot

Simultaneous​ ​Localization​ ​and​ ​Mapping​ ​(SLAM)​ ​can​ ​be​ ​implemented​ ​a​ ​number​ ​of​ ​ways​ ​in​ ​robotics
depending​ ​on​ ​the​ ​sensors​ ​used​ ​via​ ​various​ ​ROS​ ​packages​ ​that​ ​exist.​ ​ ​Here,​ ​you​ ​will​ ​use​ ​a​ ​ROS​ ​SLAM
package​ ​and​ ​simulated​ ​sensor​ ​data​ ​to​ ​create​ ​an​ ​agent​ ​that​ ​can​ ​both​ ​map​ ​the​ ​world​ ​around​ ​it​ ​and
localize​ ​within​ ​it.

Course​ ​11:​ ​Reinforcement​ ​Learning​ ​for​ ​Robotics

Begin​ ​by​ ​learning​ ​how​ ​to​ ​build​ ​a​ ​basic​ ​end-to-end​ ​reinforcement​ ​learning​ ​agent,​ ​termed​ ​a​ ​deep​ ​Q-network
(DQN).​ ​Then,​ ​enhance​ ​it​ ​to​ ​create​ ​a​ ​more​ ​complex​ ​agent​ ​that​ ​can​ ​pick​ ​and​ ​place​ ​from​ ​visual​ ​input.

Lesson Title Description

 1 RL​ ​Basics Nvidia​ ​presents​ ​the​ ​basics​ ​of​ ​RL​ ​with​ ​DQN​ ​agents​ ​and​ ​the
OpenAI​ ​Gym​ ​to​ ​train​ ​an​ ​agent​ ​to​ ​play​ ​a​ ​simple​ ​game.

2 RL​ ​Manipulation Building​ ​on​ ​the​ ​same​ ​RL​ ​engine,​ ​adapt​ ​it​ ​to​ ​solve​ ​robotics
arm​ ​manipulation​ ​problems​ ​in​ ​a​ ​Gazebo​ ​simulation
environment.

3 Project:​ ​RL​ ​Pick​ ​and​ ​Place Set​ ​up​ ​a​ ​pick​ ​and​ ​place​ ​project​ ​using​ ​the​ ​RL​ ​engine.
Robot Compare​ ​this​ ​more​ ​general​ ​method​ ​of​ ​learning​ ​to
manipulate​ ​a​ ​robotic​ ​arm​ ​with​ ​more​ ​traditional​ ​methods.

Project​ ​8:​ ​RL​ ​Pick​ ​and​ ​Place​ ​Robot

Build​ ​an​ ​RL​ ​agent​ ​to​ ​pick,​ ​grip,​ ​stack,​ ​and​ ​pack,​ ​using​ ​a​ ​manipulator​ ​arm.

Course​ ​12:​ ​Path​ ​Planning​ ​and​ ​Navigation

​ ​Extend​ ​your​ ​RL​ ​Engine​ ​with​ ​more​ ​advanced​ ​techniques​ ​for​ ​navigation,​ ​and​ ​compare​ ​these​ ​with​ ​classic
approaches.​ ​ ​Finally,​ ​combine​ ​SLAM​ ​and​ ​navigation​ ​into​ ​a​ ​single​ ​comprehensive​ ​project!

Lesson Title Description

1 Classic​ ​Path​ ​Planning Introduction​ ​to​ ​classic​ ​2D​ ​and​ ​3D​ ​path​ ​planning​ ​and​ ​the
ROS​ ​modules​ ​that​ ​implement​ ​them.

2 Lab:​ ​Path​ ​Planning Path​ ​planning​ ​with​ ​classic​ ​algorithms.

3 Navigation​ ​with​ ​RL Extending​ ​the​ ​RL​ ​engine​ ​to​ ​navigation.

4 Comparisons​ ​of​ ​Classic​ ​vs In​ ​this​ ​changing​ ​field,​ ​it's​ ​important​ ​to​ ​understand​ ​the
Deep​ ​Learning​ ​Approaches advantages​ ​and​ ​disadvantages​ ​of​ ​different​ ​approaches​ ​to
in​ ​Robotics robotics​ ​problems.​ ​ ​Sometimes​ ​the​ ​best​ ​solution​ ​is​ ​a
combination​ ​of​ ​solutions.

5 Project:​ ​Home​ ​Service Set​ ​up​ ​a​ ​project​ ​that​ ​combines​ ​advanced​ ​RL​ ​navigation
Robot with​ ​SLAM

Project​ ​9:​ ​Home​ ​Service​ ​Robot

You've​ ​already​ ​used​ ​probabilistic​ ​methods​ ​with​ ​SLAM​ ​to​ ​map​ ​and​ ​localize​ ​a​ ​robot.​ ​ ​You've​ ​also
designed​ ​deep​ ​RL​ ​engines​ ​to​ ​solve​ ​end-to-end​ ​sense-to-action​ ​problems,​ ​which​ ​can​ ​now​ ​be​ ​applied​ ​to
navigation.​ ​ ​In​ ​this​ ​project,​ ​you'll​ ​combine​ ​both​ ​AI​ ​paradigms​ ​to​ ​build​ ​a​ ​home​ ​service​ ​robot​ ​that​ ​can
map,​ ​localize,​ ​and​ ​navigate​ ​to​ ​perform​ ​household​ ​tasks,​ ​moving​ ​from​ ​one​ ​room​ ​to​ ​another
autonomously.