mobi le servi ce roboti cs roboti cs
play

MOBI LE & SERVI CE ROBOTI CS ROBOTI CS CFI DV CA 01 - PowerPoint PPT Presentation

CY 02CFI C MOBI LE & SERVI CE ROBOTI CS ROBOTI CS CFI DV CA 01 Supervision and control OBOTI C Basilio Bona RO DAUIN Politecnico di Torino Basilio Bona DAUI N Politecnico di Torino 001/ 1 Supervision and Control a


  1. CY 02CFI C MOBI LE & SERVI CE ROBOTI CS ROBOTI CS CFI DV CA – 01 Supervision and control OBOTI C Basilio Bona RO DAUIN – Politecnico di Torino Basilio Bona – DAUI N – Politecnico di Torino 001/ 1

  2. Supervision and Control a priori knowledge Task/mission commands Position CY Global map p 02CFI C Path planning Path planning Localization L li ti reasoning Map Building CFI DV Data Data Data Data ol on contro CA – 01 rception treatment treatment commands data data OBOTI C Per Motio Actuators Sensors RO Environment Environment Basilio Bona – DAUI N – Politecnico di Torino 001/ 2

  3. Supervision and Control CY Position 02CFI C Path planning Global map Localization Reasoning Map Building CFI DV Local map Path World model CA – 01 Motion Perception control control OBOTI C Environment Environment RO Basilio Bona – DAUI N – Politecnico di Torino 001/ 3

  4. Control Strategies � Structure of the control loop St t f th t l l � World changes dynamically CY � A compact model of the world does not exist A compact model of the world does not exist 02CFI C � There are many sources of uncertainty, both in the world and in the robot � Two possible approaches T ibl h CFI DV – Classic AI – deliberative model � Complete modeling (model based method) � Complete modeling (model-based method) CA – 01 � Function based � Horizontal decomposition OBOTI C � Top-down approach – Modern AI – reactive model � No world model: behavior-based � No world model: behavior-based RO � Vertical decomposition � Bottom-up approach Basilio Bona – DAUI N – Politecnico di Torino 001/ 4

  5. Control Strategies DELI BERATI VE REACTI VE Model-based Behavior-based CY 02CFI C Purely symbolic Purely symbolic Reflexive Reflexive Speed of response CFI DV Predictive capabilities CA – 01 Depends on accurate world models OBOTI C • Depends on the world representation • Representation-free • Slow response • Real-time response RO • High level intelligence (cognition) High level intelligence (cognition) • Low level intelligence (stimulus-response) Low level intelligence (stimulus response) • Variable latency • Fast and easy computation Basilio Bona – DAUI N – Politecnico di Torino 001/ 5

  6. Control Characteristics Sense Sense – Plan – Act Plan Act Subsumption/Reactive model Subsumption/Reactive model This architecture may prevent a fast http://ai.eecs.umich.edu/cogarch0/subsump/ CY and timely response and timely response 02CFI C CFI DV Task 1 sense ch approac CA – 01 Task 2 use model Task 3 Task 3 Vertical a OBOTI C plan Task 4 RO V act Task 5 Basilio Bona – DAUI N – Politecnico di Torino 001/ 6

  7. Model-Based Approach � Complete modeling of the world � Each block is a computed function CY 02CFI C � Vertical decomposition and nested-embodiment of functions CFI DV An example sensors Perception CA – 01 Localization - Map building OBOTI C Cognitive planning RO Motion control actuators Basilio Bona – DAUI N – Politecnico di Torino 001/ 7

  8. Model-Based Approach A A second example: nested embodiment d l t d b di t CY 02CFI C High level mission High level mission Service Task CFI DV Elemental move Motion primitives Servo Servo CA – 01 OBOTI C RO Basilio Bona – DAUI N – Politecnico di Torino 001/ 8

  9. Model-Based Approach A third example: nested embodiment Planner GOAL RECOGNITION CY GLOBAL PATH PLANNING 02CFI C Navigator SUB-GOAL FORMULATION LOCAL PATH PLANNING CFI DV Pilot TARGET GENERATOR DYNAMIC PATH PLANNING CA – 01 Path monitor TARGET LOCATION PATH CORRECTION/OBSTACLE AVOIDANCE OBOTI C Controller COMMANDS RO Low level control Low level control SENSORS ACTUATORS Basilio Bona – DAUI N – Politecnico di Torino 001/ 9

  10. Behavior-Based Approach � Reactive systems R ti t � Reflexive behavior CY � Perception-action � Perception action 02CFI C � Subsumption CFI DV ROBOT ROBOT CA – 01 Perception 1 Action 1 Action 2 Perception 2 p OBOTI C RO WORLD WORLD Basilio Bona – DAUI N – Politecnico di Torino 001/ 10

  11. Behavior-Based Approach Rodney Brooks is the father of this approach: Rodney Brooks is the father of this approach: Some of his key sentences CY 02CFI C � Complex behavior need not necessarily be the product of a complex control system � CFI DV Intelligence is in the eye of the observer ll h f h b � The world is its best model � � Simplicity is a virtue Simplicity is a virtue CA – 01 � Robots should be cheap � Robustness in the presence of noisy or failing sensors is a design goal OBOTI C � Planning is just a way of avoiding figuring out what to do next � All onboard computation is important RO � S Systems should be built incrementally t h ld b b ilt i t ll � No representation. No calibration. No complex computers. No high band communication Basilio Bona – DAUI N – Politecnico di Torino 001/ 11

  12. Behavior-Based Approach � No model is necessary � Horizontal decomposition CY 02CFI C � Coordination + Priority = Fusion C di ti P i it F i � Biomimesis = observe and copy animal behavior � Subsumption Subsumption CFI DV � Embodiment CA – 01 OBOTI C RO Basilio Bona – DAUI N – Politecnico di Torino 001/ 12

  13. Subsum ption � The subsumption architecture was originally � The subsumption architecture was originally proposed by Brooks [ 1986] . CY 02CFI C � The subsumption (or 'Brooksian') architecture is based on the synergy between sensation and actuation in lower animals such as insects actuation in lower animals such as insects. CFI DV � Brooks argues that instead of building complex g g p CA – 01 agents in simple worlds, we should follow the evolutionary path and start building simple agents in OBOTI C the real, complex and unpredictable world. h l l d di bl ld � From this argument, a number of key features of From this argument, a number of key features of RO subsumption result: Basilio Bona – DAUI N – Politecnico di Torino 001/ 13

  14. Subsum ption 1. No explicit knowledge representation is used. Brooks 1 No explicit knowledge representation is used Brooks often refers to this as “ The world is its own best model ” CY 02CFI C 2 2. Behavior is distributed rather than centralized. h d b d h h l d 3. Response to stimuli is reflexive – the perception-action p p p CFI DV sequence is not modulated by cognitive deliberation 4. The agents are organized in a bottom-up fashion. Thus, 4 The agents are organized in a bottom up fashion Thus CA – 01 complex behaviors are fashioned from the combination of simpler, underlying ones OBOTI C 5. Individual agents are inexpensive, allowing a domain to be populated by many simple agents rather than a few be populated by many simple agents rather than a few RO complex ones. These simple agents individually consume little resources (such as power) and are expendable, making the investment in each agent minimal ki th i t t i h t i i l Basilio Bona – DAUI N – Politecnico di Torino 001/ 14

  15. Subsum ption � Several extensions (Mataric, 1992) have been proposed to pure reactive subsumption systems. CY 02CFI C � These extensions are known as behavior-based architectures. architectures. CFI DV � Capabilities of behavior-based systems include landmark detection and map building learning to landmark detection and map building, learning to CA – 01 walk, collective behaviors with homogeneous agents, group learning with homogeneous agents, and g p g g g , OBOTI C heterogeneous agents . RO Basilio Bona – DAUI N – Politecnico di Torino 001/ 15

  16. Em bodim ent � To embody (verb) = manifest or personify in concrete T b d ( b) if t if i t form; incarnate; incorporate, unite into one body CY � Em bodim ent is the way in which human (or any other � Em bodim ent is the way in which human (or any other 02CFI C animal) psychology arises from the brain & body physiology physiology CFI DV � It is specifically concerned with the way the adaptive function of categorization works, and how things acquire g , g q CA – 01 names � It is distinguished from developmental psychology and OBOTI C physical anthropology by its focus on cognitive science, ontogeny, ontogenetics, chaos theory and cognitive RO notions of entropy notions of entropy – far more abstract and more reliant far more abstract and more reliant on mathematics Basilio Bona – DAUI N – Politecnico di Torino 001/ 16

  17. Em bodim ent � Embodiment theory was brought into AI by Rodney Brooks in the 1980s CY 02CFI C � Brooks and others showed that robots could be more B k d th h d th t b t ld b effective if they “thought” (planned or processed) and perceived as little as possible and perceived as little as possible CFI DV � The robot's intelligence is geared towards only handling the minimal amount of information handling the minimal amount of information CA – 01 necessary to make its behavior be appropriate and/ or as desired by its creator OBOTI C � Brooks (and others) have claimed that all autonomous agents need to be both embodied and RO situated. They claim that this is the only way to it t d Th l i th t thi i th l t achieve strong AI Basilio Bona – DAUI N – Politecnico di Torino 001/ 17

Recommend


More recommend