X011. Behavior based robots and architectures.ppt

44 slides
1.36 MB

Similar Presentations

Presentation Transcript


Behavior-based Robot Design An IntroductionLecture #2, Sept 8, 2005 RSS II Una-May O’Reilly


AgendaIntuition of BB design with an example Overview Practicalities


I. A Collecting Robot in Simulationwww.behaviorbasedprogramming.comBumperLeft PhotocellRight PhotocellRight IR DetectorLeft IR DetectorDrive wheels


QuestionsWhat task is the robot doing? Searching for pucks When it finds one, pushes it to vicinity of the light source, goes to find another Avoids or escapes from encounters with other objects


How is the robot collecting pucks?Task is decomposed as a set of simple behaviors (algorithms connecting sensors to actuation) that, when acting together, produce the overall activity


Collection Task Behavior NetworkEscapeDark-pushAnti-mothAvoidHomeCruiseBump forcePhotocellsIR detectorsArbiterMotor ControllerLeft MotorRight MotorSensingIntelligenceActuation


A Collecting Robot in SimulationThe robots in BSim are circular differential drive robots with a bumper, two IR proximity sensors, two photo sensors and wheel encoders. The photo and IR sensors face diagonally from the front of the robot at 45 degree angles. Each robot supports a simple, yet powerful, behavior-based programming system which includes a set of primitive behaviors and a priority list arbiter. A robot's program is called a task. A task is a prioritized list of behaviors which all simultaneously compete to control the robot. The arbiter chooses which behavior is successful. You can program each robot by configuring a set of behaviors, prioritizing the behaviors for the arbiter, and then loading the behaviors into the robot.


Collection Behaviors Cruise: drives the wheels at constant speeds. The behavior can try to drive the wheels at any speed, positive or negative, but the robot speed will max out at +/- 255. Home: tries to drive the robot toward a light source. It uses a proportional controller to home on a light source whenever the robot’s photo sensors see light. The robot homes on the light by pivoting in the direction of the light and then moving forward a step. The robot determines the direction to the light by calculating the difference between the two photo sensor measurements.. Avoid: Moves robot forward and left if the right proximity sensor is on, or forward and right is the left proximity sensor is on (if gain is positive). With a negative gain (in collection task) it goes toward an obstacle (eg a puck or wall)


Collection BehaviorsEscape:a ballistic behavior triggered whenever the robot bumps into something. The behavior is performed in three steps: backup for a specified amount of time, spin a certain angle, and go forward for a specified amount of time. Anti-Moth: a ballistic behavior that triggers whenever the total light intensity measured by a photocell exceeds a threshold Dark-push: a ballistic behavior. It triggers whenever the robot tries to push something when no light is visible.


Collection Task Behavior NetworkEscapeDark-pushAnti-mothAvoidHomeCruiseBump forcePhotocellsIR detectorsArbiterMotor ControllerLeft MotorRight MotorSensingIntelligenceActuationBacks up from wallsPrevents pushing in wrong directionDrop puck at lightFind and push a puckOrient to light source


Things to NoticeThere’s no explicit FindPuck behavior No PushPuck behavior No DropPuck behavior These emerge from the interaction of the more primitive behaviors System behavior is not deterministic, but has random components Overall behavior is robust - ultimately collects pucks No representation of the world and no state


II. Overview: Artificial CreaturesContrast between good old fashioned Artificial Intelligence (GOFAI) and behavior-based AI GOFAI: Thought experiments on the nature of “intelligence” in creatures with bodies BB-AI draws inspiration from neurobiology, ethology, psychophysics, and sociology


Good Old Fashioned AI: GOFAIintelligence --look for essence study that generalize backthe program


Marvin Minsky: Society of Mind2.5 EASY THINGS ARE HARD In attempting to make our robot work, we found that many everyday problems were much more complicated than the sorts of problems, puzzles, and games adults consider hard.


Where Did Evolution Spend Its Time?


Creature, or Behavior-Based, AIcreatures --live in messy worlds performance relative to the world intelligence (emerges) on this substratethe creature


Methodologies Compared


Embrace HubrisWhile it turns out that biological systems often use simple tricks to accomplish their goals, they are often more subtle than human engineers with all their mathematics and power tools may think they are.




Contrast: Thinking about CreaturesSimple creatures occupy very complex worlds they are not all knowing masters of the worlds they act enough to capitalize on specific features of the world They do not have enough neurons to build full reconstructions of the world The `diameter’ of their nervous systems is very small (about six for humans)


Herbert Simon’s AntA man, viewed as a behaving system, is quite simple. The apparent complexity of his behavior over time is largely a reflection of the complexity of the environment in which he finds himself.


Embrace SituatednessThe behavior of a creature, depends on the environment in which it is embedded or situated. Creatures don’t deal with abstract descriptions, but with the “here” and “now” of their environment


Embrace EmbodimentAn embodied creature is one which has a physical body and experiences the world, at least in part, directly through the influence of the world on that body. The actions of a creature are part of a dynamic with the world and have immediate feedback on the creature’s own sensations through direct physical coupling and its consequences.


Look for EmergenceThe intelligence of the system emerges from the system’s interactions with the world and from sometimes indirect interactions between its components-- it is sometimes hard to point to one event or place within the system and say that is why some external action was manifested.


AutonomousAn autonomous (artificial) creature is one that is able to maintain a long term dynamic with its environment without intervention. Once an autonomous artificial creature is switched on, it does what is in its nature to do.


Distinguish the Observer from the RobotTerms descriptive of behavior are in the eye of the observer.

Browse More Presentations

Last Updated: 8th March 2018

Recommended PPTs