Robot behaviour and control A robot can be defined as an intelligent - - PowerPoint PPT Presentation

www.biorobotics.ttu.ee

Robot behaviour and control

• A robot can be defined as an intelligent link

between perception and action

Environment Robot Perception Action

www.biorobotics.ttu.ee

For a mobile robot

Raw data Environment Model Local Map "Position" Global Map Actuator Commands Sensing Acting Information Extraction Path Execution Cognition Path Planning Knowledge, Data Base Mission Commands Path Real World Environment Localization Map Building

Motion Control Perception

www.biorobotics.ttu.ee

Reacting to the world

• Deliberate

– Creating a world model – Making a plan – Execution and monitoring the outcomes

• Reflexive behaviour

– Immediate reaction to stimuli

• Hybrid planners

– Reflexive locally, deliberate globally

www.biorobotics.ttu.ee

Behaviour-based robots

• Robot behaviours are defined as sensor-

motor pairs

• Connect stimuli to reactions without

intermediate deliberate thinking (“ the world is its own best model”)

• Reactive robots have no internal

representations

• Reactive robots have no memory

www.biorobotics.ttu.ee

Braitenberg vehicles

• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Braitenberg vehicles

• R. Arkin, “Behavior-based robotics”

http://people.cs.uchicago.edu/~wiseman/vehicles

www.biorobotics.ttu.ee

Common behaviours

• Moving towards a light source
• Moving away from a light source
• Obstacle avoidance
• Wall following
• Line following
• Wandering
• Maintaining stability

www.biorobotics.ttu.ee

BEAM robotics

• BEAM – Biology, Electronics, Aesthetics,

Mechanics

• Uses simple analog circuits to produce an

unusually simple design

• Analog circuits are used to simulate

biological neuron behaviours

www.biorobotics.ttu.ee

Design of behaviour based robots

• What behaviours do we need for the robot

to fulfil its tasks?

• Create sensor-motor pairs. What are the

stimuli triggering a behaviour and what are the reactions?

• Modular design – combine behaviours to

get more complicated behaviours

www.biorobotics.ttu.ee

Emergent behaviours

• Emergent behaviour – collection of

behaviours produce results that were not explicitly programmed

• The Behavior-Based approach states that

intelligence is the result of the interaction among a set of behaviors and the environment (“ Intelligence is in the eye of the observer “).

www.biorobotics.ttu.ee

Qualities of emergent behaviours

• Robust and resilient
• Easy to design
• Fault tolerant due to the lack of central

control

• Fast due to the lack of symbolic

representation and planning

www.biorobotics.ttu.ee

Comparison of deliberate and reactive robots

www.biorobotics.ttu.ee

Subsumption architectures

• R: Brooks, 1986
• Designing reactive architectures bottom-up using

layered sets of rules

• Bottom layer behaviours are mostbasic and have

highest priority

• Top level behaviours encapture high-level

intentions

• Uppler lever can only function if lower level

behaviours are satisfied

• Minimal interaction between behaviours

www.biorobotics.ttu.ee

Swarm intelligence

• based on the collective

behavior of decentralized, self-organized systems

• Swarm consists of a small

number of simple physical robots

• desired collective behavior

emerges from the interactions between the robots and interactions of robots with the environment.

http://www.newscientist.com

www.biorobotics.ttu.ee

Designing behaviour-based

• robots. Ethological design
• The behaviours are consistent with

biological evidence

• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Designing behaviour-based

• robots. Situation-based

design

• The reaction of the

robot depends on the situation

• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Designing behaviour-based

• robots. Experimental design
• Bottom-up
• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Representing behaviours

• Functional

representation

• b(s) = r

– b – behaviour – s – stimulus – r – reaction

www.biorobotics.ttu.ee

Representing behaviours

• Finate-state automata
• M := (Q, δ, q0

, F)

– Q – set of behaviours – δ – transfer function – q0 – initial behaviour – F – set of states

δ q Input δ(q, input) b a b

M = {{b}, δ, b, {b}}

www.biorobotics.ttu.ee

Representing behaviours

• Navigating to the

lecture room with FSA diagram

www.biorobotics.ttu.ee

Coding reactions

• Magnitude of reactions
• Direction of reactions
• Behaviour (S, R, β)

– S – stimulus – R – reaction – β: S R

www.biorobotics.ttu.ee

Discrete coding

• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Continuous coding

• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Coordination of behaviours. Priority based coordination (subsumption)

• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Action selection

• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Action coordination

• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Action coordination of

• bstacle avoidance and goal

following

• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Action coordination of

• bstacle avoidance and goal

following

• R. Arkin, “Behavior-based robotics”

www.biorobotics.ttu.ee

Home assignment

Mandatory: Braitenberg vehicles. Use the Braitenberg vehicle simulator at http://www.ai.rug.nl/~gert/applets/braitenbergJRE/ Observe the behaviour of the vehicles in response to the light inc ase of different placements of light sources. Voluntary: implement obstacle avoidance Brainteberg vehicles behaviour in the robot simulator

• R. Arkin, “Behavior-based robotics”