Behavior-based Control
Part 2: Robotic Implementation
CIS548
[Note: References to Parker are from Sep. 10 lecture notes].
Reactive vs. deliverative control:
Parker, p. 6: “A reactive robotic system tightly couples perception to action without the use of intervening abstract representations or time history”
Parker, p. 7: “Use of explicit abstract representational knowledge is avoidedin the generation of a response”
[Note: Recall Brooks/Kirsh debate from previous lecture]
Reactive vs. deliberative – pp. 9-11
[Note: We will design reactively at lowest levels, yet deliberatively at highest levels]
3 design methods:
Parker, pp. 13-24:
Ethological
Situated activity [Agre, Chapman, Pengi]; universal planning
Design-test-revise cycle
Types of robotic behaviors:
Inspired by Parker, pp. 25-27, but amended
– Exploration
o mapping
o searching/seeking
– Movement
o toward destination
o heading-based
o taxis-based
§ toward
§ away from
o path following
§ line following
§ hallway/roadway following
o aversive
o elusive
– goal-oriented
o task-oriented
o appetitive
o protective
– postural
o balance
o stability
– social/cooperative
o human-interactive
o cooperative goal achievement
o flocking/herding
– teleautonomous (human operator coordinated)
– perceptual
o saccades [See immediately below]
o etc.
– non-wheel locomotive
o walking
§ multi-legged
§ two legged
o gait control
– manipulator-specific
o reaching
o grasping
o enveloping
Behavioral problem example – eye motion:
From wikipedia: http://en.wikipedia.org/wiki/Saccade
“Saccadic eye motion
In regards to the eye, saccades are quick, simultaneous movements of both eyes in the same direction [1]. Initiated by the frontal lobe of the brain (Brodmann area 8), saccades serve as a mechanism for fixation, refixation, rapid eye movements and the fast phase of optokinetic nystagmus [1].
Humans and other animals do not look at a scene in a steady way. Instead, the eyes move around, locating interesting parts of the scene and building up a mental 'map' corresponding to the scene. One reason for saccades of the human eye is that only the central part of the retina, the fovea, has a high concentration of color sensitive photoreceptor cells called cone cells. The rest of the retina is mainly made up of monochrome photoreceptor cell called rod cells, which are especially good for motion detection. Consequently, the fovea makes up the high-resolution central part the of human retina.
By moving the eye so that small parts of a scene can be sensed with greater resolution, body resources can be used more efficiently.
The dynamics of saccadic eye motion give insight into the complexity of the mechanism that controls the motion of the eye. The saccade is the fastest movement of an external part of the human body. The peak angular speed of the eye during a saccade reaches up to 1000 degrees per second. Saccades last from about 20 to 200 milliseconds.
The duration of a saccade depends on its amplitude. The amplitude of a saccade is the angular distance that the eye needs to travel during the movement. For amplitudes up to about 60 degrees, the duration of a saccade linearly depends on the amplitude. In that range, the peak velocity of a saccade linearly depends on the amplitude. In saccades larger than 60 degrees, the peak velocity remains constant at the maximum velocity attainable by the eye. Thus, the duration of these large saccades is no longer linearly dependent on the amplitude.
In addition to the kind of saccades described above, the human eye is in a constant state of vibration, oscillating back and forth at a rate of about 60 per second. These microsaccades are tiny movements, roughly 20 arcseconds in excursion and are completely imperceptible under normal circumstances. They serve to refresh the image being cast onto the rod cells and cone cells at the back of the eye. Without microsaccades, staring fixedly at something would cause the vision to cease after a few seconds since rods and cones only respond to a change in luminance.”
Encoding behaviors – Expression:
Parker, pp. 28-35
p. 28::
“Expression of Behaviors – 3 methods
– Stimulus-Response Diagrams
• Useful for graphic representations of specific behavioral configurations
– Functional Notation
• Useful for clarity in design of systems
– Finite State Acceptor Diagrams
• Useful when temporal sequencing of behaviors is required”
Encoding behaviors – SR mapping:
Parker, pp. 36-47
“Behavioral encoding:
creates functional mapping from the stimulus plane to the motor plane
Behavior expressed as a triple: (S, R, β )
where:
S denotes domain of all interpretable stimuli
R denotes range of possible responses
Β denotes mapping β: S → R”
“Each individual stimulus, or percepts (s εS) represented as tuple (p,λ)
where:
p = perceptual class (or type)
λ= property of strength”
“When a stimulus threshold is in effect, we have:
β: (p,λ) |� {for all λ< τ then r= [0, 0, 0, 0, 0, 0] \* no response
*\else r= arbitrary function} \* response *\”
“Three Categories of Behavioral Mappings
1. Null: stimulus produces no motor response
2. Discrete: stimulus produces a response from an enumerable set of
prescribed choices
E.g.: turn-right, go-straight, stop, travel-at-speed-5
3. Continuous: stimulus produces a motor response that is continuous
over R’s range”