The advent of Mobile Robotics changed the definition of robotics and brought in some very interesting technologies paving the way for cutting edge sciences like AI, Behaviour Based Systems, etc
3. Where did it begin?
1962 – General Motors
General Motors install a Unimate as
the world's first robot on a
production line
Unimate then launched PUMA, the
world's first 6 axis articulated robot
which led to a new generation of
flexible manufacturing systems and
the core technology in robotics
4. Where did it begin?
1966 – 1972: Stanford Research Lab
“Shakey” from Stanford was the world's first mobile
robot to reason about its actions
Endowed with a limited ability to perceive and model
its environment, Shakey could perform tasks that
required planning, route-finding, and the rearranging
of simple objects.
In short, Shakey was the path maker to todays
intelligent robots
7. Adding Mobility – Mobile Platforms
Mobile Platforms can move freely & hence have limitless operational area
These platforms however cannot manipulate objects themselves
Typical tasks involve surveillance, cleaning, monitoring & analysis – no handling
The platforms have different mechanisms to move – wheels, legs, wings, even jets!
Has the ability to move around human presence – hence must be safe
Safety measures demand embodied intelligence – hence the rise of AI
Operations like obstacle avoidance, map analysis and self awareness are all parts of
the overall AI of the system
8. Mobile Platforms – Drive Concepts
Differential Drive
Two powered wheels
Other wheels are passive and free (castor)
Fast moving, but limited mobility
2 Degrees of freedom
9. Mobile Platforms – Drive Concepts
Synchro Drive
All wheels are powered
Steering is also powered but synchronized
All wheels steer the same way simultaneously
Fast moving, but limited mobility
2 Degrees of freedom
Advantage over DD:
Can move at an angle to heading
10. Mobile Platforms – Drive Concepts
Omni Drive
Three / Four powered wheels
Wheels are specially designed
Fast moving, high mobility
3 degrees of freedom – can move at any
heading and turn at the same time
11. Mobile Platform – DD Kinematics
Notations:
Vr, Vl are right and left wheel velocities
l/2 is the wheel separation
W is the angular velocity of the robot
(the rate at which the robot is rotating
about the vertical axis)
ICC is the Instantaneous center of
curvature
R is the distance of the robot base to
ICC
(x, y) is the robot position
θ is the robot orientation
12. Mobile Platform – DD Kinematics
From basic equations of motion
Vr = (R + l/2).W
Vl = (R – l/2).W
Solving, we get
R = l/2 (Vl + Vr) / (Vr – Vl)
W = (Vr – Vl) / l
If Vr = Vl: We get R as infinity and the robot travels straight
If Vr = -Vl: R becomes 0 (zero) and the robot turns on its ICC or the base mid point
For other values, the robot will steer left or right depending on the speed difference. The
value of R can be calculated
13. Mobile Platform – DD Kinematics
Question.....
For the robot ICC to be located under
the left wheel, what should be the wheel
velocities?
14. Mobile Platform – DD Kinematics
Under the left wheel means that
R = l/2
Thus substituting in
R = l/2 (Vl + Vr) / (Vr – Vl)
l/2 = l/2 (Vl + Vr) / (Vr – Vl)
Vr - Vl = Vl + Vr
ie 2Vl = 0 or Vl = 0
Thus Vl = 0 and for any Vr, the
condition will be met
16. Where am I? Robot Localization
If you see this.... where will you be?
17. Where am I? Robot Localization
If you see this.... where will you be?
18. Where am I? Robot Localization
If you see this.... where will you be?
19. Where am I? Robot Localization
Any particular object which can be uniqueliy identified
and mapped to a location is called a LANDMARK
Robots work on Landmarks for Localization using many different
mathematical models
The simplest one being triangulation
P1
Landmark1
R1
Question:
How many Landmarks are needed to
P(?) uniquely get the robot location P
P2
R2 The robot is able to identify the
Landmark2 Robot landmark, landmark position & its own
distance Rx from the landmark
20. Where am I? Robot Localization
Landmark1
Landmark2
21. Where am I? Robot Localization
Landmark1
Landmark2
Two Landmarks can give you a
“false positive”
22. Where am I? Robot Localization
Landmark1
Landmark2
A third landmark will eliminate the
wrong location
23. (R)Evolutionary Learning Systems
We have seen that Inverse Kinematics is a complex task
So can we really program a robot to walk?
2 DOF = 2 solutions
4 DOF = 8 solutions
6 DOF = 16 solutions
.........
24 DOF = ???
24. (R)Evolutionary Learning Systems
Evolutionary techniques are used to make the robot “learn”
- learn to adapt to environment
- learn to avoid obstacles
- learn to navigate
- or even learn to walk!
Video of DFKI Walking
25. What is Evolution Algorithms
Follows darwin principle – survival of the fittest
Case Study: Walking Robots
a. Take a number (100+) of robot programs which try to make the robot walk
b. Run each of the programs on the given robot
c. EVALUATION: Analyse the best programs
– the best will make the robot walk the farthest / fastest
d. SELECTION: Choose the best 10 – 20 programs
d. MUTATION: Create (automatically) programs which are similar to the best
but slightly modified
e. Run the new batch of programs again
f. Repeat the Evaluation, Selection & Mutation
At the end we will get a few programs which will
make the robot walk fast, stable and far!
26. So what is the “New Age” ?
Adaptable
This walking robot developed at Fraunhofer is able
to walk even if one of its legs are damaged!
27. So what is the “New Age” ?
Interactive
Pleo – the robot dinosaur interacts with his master
and actually develops his own personality
28. So what is the “New Age” ?
Safe
Care-o-bot (left) and Justin (right) are designed to be safe in
presence of humans. They are interactive and adaptable too!