Robotics week2

Question 1

What are the five main components that make up a robot?

Answer 1

1. Body : materials and morphology.
2. Sensor: for perception.
3. Effector: for action.
4. Control mechanism (brain): intelligence, cognition, responsible for behaviour.
5. Energy/power source: e.g., electricity, pneumatic (gas), hydraulic (liquid).

Question 2

Why is “degrees of freedom” (DoF) important for robot morphology?

Answer 2

1. DoF equals the number of independent parameters defining a robot’s configuration.
2. A rigid object in 3D space has six DoF: three for position and three for orientation.
3. A manipulator needs at least six joints (DoF) to position its end-effector arbitrarily; additional DoF (redundancy) can optimize tasks (e.g., energy minimization, obstacle avoidance).
4. Example: the human arm has seven DoF; a robotic snake may have many DoF.

Question 3

In the simplified pictographic representation, what four interactions connect the robot to its environment?

Answer 3

1. Sensing:
   obtaining information from the environment.
2. Actuation:
   generating movement or signals.
3. Body:
   the physical form through which sensing and actuation occur.
4. Control:
   processing sensed data and commanding actuator responses.

Question 4

List at least four examples of sensors used in robotics and distinguish between low-level and high-level percepts.

Answer 4

• Sensors: cameras, microphones, force/pressure sensors, infrared, ultrasound, LIDAR, compass, gyroscope, GPS, etc.
• Low-level percepts: raw measurements such as light intensity, colour, sound level, temperature, proximity, contact, tilt, acceleration, voltage, current.
• High-level percepts: interpreted data such as object identities, people, scenes.
• Abstract percepts: intentions, meanings, affective states (inferred).

Question 5

What is the difference between external perception and internal proprioception? Give an example of each

Answer 5

External perception: sensing information about the environment (e.g., a LIDAR measuring obstacle distances).

Internal proprioception: sensing the robot’s own state (e.g., joint angles of a manipulator, battery voltage)

Question 6

Explain “direct perception” versus “inference” in robotic sensing.

Answer 6

Direct perception: extracting immediate information without intermediate models (e.g., using optic flow to estimate time-to-target).

Inference: constructing hypotheses and testing to derive higher-level meaning (e.g., recognizing an object by segmenting an image and matching to a database).

Question 7

What are the advantages and disadvantages of a robot sensing its own actions during interaction?

Answer 7

Advantage:

The robot can verify if it is performing as intended (e.g., using vision to confirm it has grasped an object).

Disadvantage:

Sensor readings may be corrupted by its own actuation (e.g., motor noise masking important sound cues)

Question 8

Define “Cybernetics” and name its three combined disciplines, as well as one example of an early cybernetic robot.

Answer 8

Cybernetics :
the study of control and communication in animals and machines.

1. Combines control theory,
2. Information science
3. Biology.

Example: Grey Walter’s Machina Speculatrix, an early robotic turtle that exhibited simple goal-seeking and obstacle-avoiding behaviours.

Question 9

Describe Hans Moravec’s Stanford Cart (1979) and its key limitations

Answer 9

A mobile robot equipped with TV cameras that captured images, planned a path, and move in 1 m increment

Limitation : Each move required 10-15 minutes of processing ( takes too long to explore the whole room )

Question 10

What are two main problems associated with the Sense-Plan-Act (SPA) approach

Answer 10

1. Closed-world assumption: Hard to model every detail, world models become huge and quickly outdated
2. Slowness :
   Inability to react in real time, making it impractical for dynamic environments

Question 11

What are Braitenberg’s Vehicles, and what do they illustrate about control?

Answer 11

: Conceptual “Thought experiment” robots comprised of simple sensor-motor connections

Theyn demonstrate that complex-seeming behaviours can arise from minimal control architectures

Question 12

Give two specific examples of Braitenberg Vehicle behaviours

Answer 12

Vehicle 1. Approach : 빛 센서는 대응하는 모터를 작동시킴, 한쪽 빛이 강해지면 빛이 강한 쪽으로 회전

Vehicle 2a. Coward : 빛 센서 반대로 모터를 작동시킴 -> 빛 반대로 도망감

Vehicle 2b Aggression : 센서 반대로 모터를 작동시키지만 밝은 빛을 향해 돌진함

Vehicle 3a Love : 센서는 반대쪽 모터를 억제, 빛이 앞쪽에 있으면 모터 모두 속도가 느려져 마치 빛 주위에 머무르는것 처럼 보임

Vehicle 3b Explorer : 센서 억제 방식이 다르게 배치되어있어서 더 복잡하게 방황

Question 13

What is the “subsumtion architecture” in the context of BBR?

Answer 13

Subsumption architecture:
a layered control structure where lower-level behaviours subsume(포섭하다) or preempt (선점하다) higher-level ones

Question 13

What is Behaviour-Based Robotics?

Answer 13

단순한 감각-행동 매핑 (sensory - motor coupling) 을 통해 자율적으로 동작하도록 설계하는 접근 방식임

현재 센서 입력(라이다, 카메라, 접촉 센서, 광센서) 기반으로 실시간으로 행동을 결정.

여러개의 behaviour 모듈이 layered(병렬)로 동작하며 이들간에 우선순위 관계를 정의하여 복합적인 동작을 만들어 낸다.

Question 14

Pros and Cons for Behaviour-based-robotics

Answer 14

장점 : high responsivenss, robostness, scalability

단점: 최적 경로 미보장, 복잡한 행동 조합의 어려움, 장거리 상황 인식의 한

Question 15

Contrast the “deliberative” (Sense-Plan-Act) approach with the “reactive” (subsumption) approach

Answer 15

SPA suited for structured, static environments, builds a detailed internal model

Subsumption suited for unstructured environments and planning in favor of layered reactive behaviors that directly “sense -> act”

Question 16

Braitenberg’s ‘vehicles’ illustrate concepts of which robotic paradigm?

Answer 16

Reactive robotics

Question 17

What is the key feature of behaviour based robotics

Answer 17

Emphasis on simple behaviours leading to complex action

Question 18

What is the central idea behind the SPA architecture?

Answer 18

Sensing informs planning which then results in action

Question 19

How does a deliberative AI robot generally handle unexpected obstacles during its operation

Answer 19

By re-evaluating its plans and adjusting its actions accordingly

Question 20

What is the primary focus of cybernetics as a field of study?

Answer 20

The study of communication and control in living organisms and machines

Question 21

What three criteria define a robot’s autonomy
1. Actuated Mechanism
2. Autonomous
3. Programmable

Answer 21

1. Actuated Mechanism: capable of movement (real or simulated).
2. Autonomous: capable of acting without constant human intervention.
3. Programmable: capable of being reconfigured via instructions

Question 22

What central problem must be solved continuously in autonomous systems

Answer 22

Action Selection, deciding “what to do next” based on sensor inputs, internal state, and objectives

Question 23

How are “self-sustaining” and “recursive self-sustaining” systems differentiated?

Answer 23

self sustaining system -> make active contribution to their own persistence but not maintain conditions for persistence

Recursive self-sustaining system -> actively maintain conditions for persistence and adapt self-maintenance processes based on environmental conditions

Question 24

What does autopoiesis imply for a robot's design?

Answer 24

Autopoiesis implies the robot has a self-sustaining body that can autonomously maintain its physical integrity and energy.

Question 25

What does homeostasis imply for a robot's control?

Answer 25

Homeostasis -. system that can autonomously regulate(규제하다) internal variables (battery level, temperature) to maintain stable operation in changing environments

Question 26

What four requirements must Behaviour-Based Robotics (BBR) satisfy

Answer 26

1. Multiple goals: Possibly conflicting objectives 2. Multiple Sensors: often providing inconsistent data 3. Robustness: tolerance to component failure 4. Extensibility: ability to add new behaviours

Question 27

Describe the layered "subsumption architecture” in BBR

Answer 27

Behaviours are organized in layers, each handling a specific priority (e.g., obstacle avoidance at Level 0, wandering at Level 1, exploration at Level 2). Higher layers can suppress or override lower layers when their stimuli are relevant Each layer runs concurrently, enabling real-time reactivity without a global planner. .

Question 27

What is the primary limitation of the subsumption architecture despite its robustness(견고성)?

Answer 27

1. stimulus--response approach -> can struggle with tasks requiring long-term planning or memory. 2. scalability issues when there's too many layers to interact

Question 27

Besides subsumption, name two other BBR approaches mentioned

Answer 27

1. Finite State Represnatations : behaviours modeled as states and transitions based on sensor inputs 2. Perception-Action Schemas : sensor-motor coupling organized into higher-level schemas 3. Potential Fields : using artificial potential gradients (attractors/repulsors) for navigation

Question 28

How do potential fields work for robot navigation? and what is the common problem with potential field navigation and the mitigation?

Answer 28

Obstacles act as repulsors -> push robot away Goals act as attractors -> pull robot towards them Robot computes a local gradient vector(sum of repulsive and attractive components) and move accordingly without path planning Problem : robot can get stuck where attractive and repulsive forces balance Mitigation: introduce a "spin" or random perturbation(change a value) when trapped.

Question 28

What is morphological computing, and why is it beneficial?

Answer 28

Morphological Computing: using a robot's body to perform computations(filtering, stability) that would otherwise require contorl feedback Benefits: inherently(본질적으로) stable, robust(건장한) to perturbations, energy-efficient and reduced sensor/control delays