Upload
bibekananda-datta
View
225
Download
1
Embed Size (px)
Citation preview
8/12/2019 robotic Paradigms
1/22
1
Robot Control Paradigms
Introduction toMobile Robotics
Wolfram Burgard, Cyrill Stachniss, Maren
Bennewitz, Diego Tipaldi, Luciano Spinello
8/12/2019 robotic Paradigms
2/22
2
Classical / Hierarchical Paradigm
! 70s! Focus on automated reasoning and knowledge
representation
! STRIPS (Stanford Research Institute Problem Solver):Perfect world model, closed world assumption
! Find boxes and move them to designated position
Sense Plan Act
8/12/2019 robotic Paradigms
3/22
3
Stanford CART 73
Stanford AI Laboratory / CMU (Moravec)
8/12/2019 robotic Paradigms
4/22
4
Classical ParadigmStanford Cart
1. Take nine images of the environment, identifyinteresting points in one image, and use other
images to obtain depth estimates.
2. Integrate information into global world model.3. Correlate images with previous image set to
estimate robot motion.4. On basis of desired motion, estimated motion,
and current estimate of environment, determinedirection in which to move.
5. Execute the motion.
8/12/2019 robotic Paradigms
5/22
5
Classical Paradigm as Horizontal/Functional Decomposition
Sense Plan Act
Perception
M
odel
Plan
Ex
ecute
Moto
rControl
ActionSensing
Environment
8/12/2019 robotic Paradigms
6/22
6
Reactive / Behavior-based Paradigm
Sense Act
! No models: The world is its own, bestmodel
! Easy successes, but also limitations! Investigate biological systems
8/12/2019 robotic Paradigms
7/227
Reactive Paradigm asVertical Decomposition
Avoid obstacles
Wander
Explore
ActionSensing
Environment
8/12/2019 robotic Paradigms
8/228
Characteristics of ReactiveParadigm
! Situatedagent, robot is integral part of theworld.
! No memory, controlled by what ishappening in the world.
! Tight couplingbetween perception andaction via behaviors.
! Only local, behavior-specific sensing ispermitted (ego-centricrepresentation).
8/12/2019 robotic Paradigms
9/229
Behaviors
! are a direct mappingof sensoryinputs to a pattern of motor actionsthat are then used to achieve a task.
! serve as the basic building block forrobotics actions, and the overallbehavior of the robot is emergent.
! support good software designprinciples due to modularity.
8/12/2019 robotic Paradigms
10/2210
Subsumption Architecture
! Introduced by Rodney Brooks86.! Behaviors are networks of sensing and
acting modules (augmented finite
state machinesAFSM).! Modules are grouped into layers of
competence.
! Layers can subsumelower layers.! No internal state!
8/12/2019 robotic Paradigms
11/2211
Level 0: Avoid
Polar plot of sonars
Collide
Feel force Run away Turn
Forward
Sonar polarplot
force heading
halt
heading
encoders
8/12/2019 robotic Paradigms
12/2212
Level 1: Wander
Collide
Feel force Run away Turn
Forward
Sonar polarplot
forceheading
halt
Wander Avoidforce
heading
s
modified
heading
heading
encoders
8/12/2019 robotic Paradigms
13/2213
Level 2: Follow Corridor
Collide
Feel force Run away Turn
Forward
Sonar polarplot
force
halt
Wander Avoidforce
heading
to middle
s
modified
heading
LookStay in
middle
s
corridor
heading
Integrate
heading
encoders
distance, direction traveled
8/12/2019 robotic Paradigms
14/22
14
Potential Field Methodologies
!Treat robot as particleacting under theinfluence of a potential field
! Robot travels along the derivative of thepotential
! Field depends on obstacles, desired traveldirections and targets
! Resulting field (vector) is given by thesummation of primitive fields
! Strength of field may change with distanceto obstacle/target
8/12/2019 robotic Paradigms
15/22
15
Primitive Potential Fields
Uniform Perpendicular
Attractive Repulsive Tangential
8/12/2019 robotic Paradigms
16/22
16
Corridor Following withPotential Fields
! Level 0(collision avoidance)is done by the repulsive fields of detectedobstacles.
! Level 1(wander)adds a uniform field.! Level 2(corridor following)
replaces the wander field by three fields(two perpendicular, one uniform).
8/12/2019 robotic Paradigms
17/22
17
Characteristics of Potential Fields
! Suffer from local minima
! Backtracking!
Random motion to escape local minimum! Procedural planner s.a. wall following! Increase potential of visited regions! Avoid local minima by harmonic functions
Goal
8/12/2019 robotic Paradigms
18/22
18
Characteristics of Potential Fields
! No preference among layers! Easy to visualize! Easy to combine different fields! High update rates necessary! Parameter tuning important
8/12/2019 robotic Paradigms
19/22
19
Reactive Paradigm
!Representations?
! Good software engineering principles?! Easy to program?! Robustness?! Scalability?
8/12/2019 robotic Paradigms
20/22
20
Hybrid Deliberative/ReactiveParadigm
Sense Act
! Combines advantages of previous paradigms! World model used for planning! Closed loop, reactive control
Plan
8/12/2019 robotic Paradigms
21/22
21
Discussion
!Imagine you want your robot to performnavigation tasks, which approach would youchoose?
! What are the benefits of the behavior basedparadigm?
8/12/2019 robotic Paradigms
22/22
22
Discussion
!Imagine you want your robot to performnavigation tasks, which approach would youchoose?
! What are the benefits of the behavior basedparadigm?
! Which approaches will win in the long run?