Scribd
Upload
20 views

Autonomous Systems - Final

Uploaded by

Kristiyana
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
20 views

Autonomous Systems - Final

Uploaded by

Kristiyana
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 4

Lecture 1 exam questions

(1) Autonomous (robot) systems capable of operating:


-without human innervations
-for prolonged time periods
-in partially unknown environments

(2) Thus, we want to say that line arms are not autonomous, but the Rumba (vacuum cleaner)
is not an autonomous system because they act in fully known environment -> we DON’T
agree with him

Lecture 2 exam questions

Sensor – a device that measures some attribute of the world in a quantified way.

(3) Classifications of sensors


Exteroceptive (measures quantities from the robot environment, i.e. external state – camera,
lidar, IR, Touch sensor)
-active – emit energy to the environment to measure something – affects the
environment – example: distance sensor, laser, sonars
-passive – receive and measure energy from the environment – doesn’t affect
the environment – example: camera, microphone.

Proprioceptive (measure the values of the robot, i.e., internal state – sensor for position,
angles, battery level)

(4) Braitenberg Vehicles – autonomous

Lecture 3 exam questions

(5) Situatedness – refers to existing in a complex, dynamic and unstructured


environment that strongly affects the robot’s behavior – we want to work with
hardware to achieve this so the robot interacts with the environment – having a
physical body and performing “mindless” doesn’t mean that it Situatedness.
i.e. being situated in external environment.
(6) Control models
Open loop control – example: microwave – setting
parameters initially, parameters don’t change while
running, desired results are not guaranteed

Closed loop control – example: Braitenburg vehicle –


data from sensor influence the behavior, system uses
errors to generate action, autonomous robots use closed
loop control

(7) Inverse and forward kinematics


-picture of a robot arm and say what kinematics as exam question

Forward kinematics - given a sequence of commands, what is the final position of robotic
arm - easy to compute

Inverse kinematics - given a desired position of the robotic arm, what sequence of commands
will bring it there - hard to compute – there can be one/multiple/no solution

(8) Robot primitives: Sense, Plan, and Act


Sensing – gather information using sensors
Planning – create a world model and plan next move
Act – generate actuator commands to perform an action

Hierarchical (deliberative) paradigm


 Planning: logic/neural/rule-based engine
 Sensing: providing representation of
environment
 Acting: robots motor behavior

Hybrid deliberative/reactive paradigm

 combines hierarchical/deliberate and reactive


paradigm

 first hierarchical/deliberative - plan


 then: reactive - sense/act

Lecture 4 exam questions

(9) What is embodiment (question about soft robotics)


soft robots are designed to mimic the flexibility and adaptability of natural organisms, with
objects more effectively, soft robots are often designed to have properties such as elasticity
and deformability
(presentation)Do you think ChatGPT is an embodied agent?, Explain why? – No, because the
main indication that a an agent is embodiment is to have a psysical body. ChatGPT is an
online application.

(presentation) Do you think Pepper and Nao robots are embodied agent?, Explain Why? –
Yes, because the main indication that a an agent is embodiment is to have a psysical body and
they both have and person can interact with them in an environmment.

(10) How to build an embodied agent?


Synthetic methodology (understanding by building): scientists trying to understand ant
behavior
(a) directly mapping observed behavior onto an internal representation
(b) using this representation to control walking in a robot behavior
(c) applying a much simpler model – resemble a braitenburg architecture – to control
walking in a robot

Lecture 5 exam questions

(11) Incomplete list of learning from demonstration


Kinesthetic teaching Teleoperation Direct imitation of human behavior

(12) Q learning
Q update function – The learning rate determines to what extent newly acquired information
overrides old information – The discount factor determines the importance of future reward
Look at extreme cases given 2 hyperparameters?
Remember it is multiplicative [what happens if it is 0 or 1] ->no function, because too big
and overwritten

Lecture 6 exam questions

(13) Role of learning rate


if 0 => no learning, just remembering weights of last
input, weights will be the same

if 1=>overlearning, in case of applies will remember


exactly what input was & not adapt weight get cancelled

Lecture 7 exam questions


(14) Intro to ML
ML is concerned with the question of how to construct computer programs that automatically
improve with experience.
ETP – computer program is said to learn from experience – E, with respect to some class of
tasks – T and performance measure – P. If its performance at tasks in T, as measured by P,
improves with experience E.

(15) Why multimodal ML?


-to form a robust sensory representation
- to leverage complementary choirs of modalities

Lecture 8 exam question

(16) Robotics vs HRI


Robotics has to do with creating physical robots and HRI concerned with the way the robot
interacts with people in the social world, happens in real life and has a social component.

Feeling – subjective representation of emotions /mood – diffuse affective states that generally
last longer/ affect- underlying affective experience of an emotion or a mood

(17) Functions of emotions (examples of these)


-Coordinating behavioral responses (flight or fight)
-Communication (social bonding)
-Short-cut cognitive processing
-Facilitating storage and recall of emotions

Lecture 10 exam questions

(18) Simulation
Smulation is the process of using a computer to approximate how a dynamic system (here
a robot) evolves in time. To this end, if the computer relies on the laws of physics to
predict the behavior of the dynamic system of interest, then the simulation is called
physics based
Computer graphics:
-define position orientation, shape of object at time -1
-render
- define position orientation, shape of object at time +1

(presentation) How can you model legged locomotion (walking, running, and jumping)
without using any animations? What is the one of the physical systems to generate legged
locomotion for humanoid robots?

Lecture 11 mock exam questions - presentation

You might also like