4 a Architectures for Autonomy Given that autonomy

4 a Architectures for Autonomy Given that autonomy has a different programming style, what is it? How Much Artificial In telligence Does a Ro bot Need? Can intelligence be added in layers? Like upgrading to “pro version” or downloading “apps” as needed? © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) http: //www. gameartisans. org 1

4 a Specific Learning Objectives • Apply Levis' definition of architectures to organizing software in an intelligent robot and be familiar with how they can guide the design process Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary • Name and describe the layers (behavioral, deliberative, interface) within a canonical operational architecture of an intelligent robot in terms of the five attributes • Name the four primitives of robotic intelligence (sense, plan, act, learn) • List vulnerabilities (e. g. , where poor choices or implicit assumptions can lead to failures) in the canonical operational architecture © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 2

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Recap: Where We’ve Been… • AI researchers view robotics as an issue of increasing autonomy (doing the “right thing” in an open world), • whereas many engineering researchers view robotics as an issue of extending automation (focus on creating a closed world and guaranteeing actions). • This gives rise to very different, though not necessarily mutually exclusive, approaches especially in terms of programming. © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 3

4 a The Big Picture • AI has converged on a canonical operational architecture after years of exploration Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary – This will be the backbone of additional lectures as we delve into the systems architecture and choice for algorithms, coordination functions, etc. – Historical development is of interest because it helps identify legacy code and to see barriers that led to the convergence (separate lecture) • Operational architectures have advantages – S/W engineering principles of abstraction, modularity – Semiformal design specification and completeness- master checklist, did you remember all the pieces – Can reveal (some) fundamental vulnerabilities © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 4

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary GIVEN THAT AUTONOMY HAS A DIFFERENT PROGRAMMING STYLE, WHAT IS IT? Via… • Architectures, specifically operational architecture • Biological operational architecture • Recast as a Process • Recast as a Function • Ramifications © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 5

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Organizing Software • Overall style of design or organization is called an “architecture” – provides a principled way of organizing a control system. However, in addition to providing structure, it imposes constraints on the way the control problem can be solved [Mataric] – describes a set of architectural components and how they interact [Dean & Wellman] © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 6

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Types of architectures [Levis, George Mason University] • operational architecture: describes what the systems does at a high level, not how it does it • systems architecture: describes how a system works in terms of major subsystems • technical architecture: describes how a system works in terms of implementation details, language © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 7

4 a [Levis, George Mason University] Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary • operational architecture: describes what the systems does at a high level, not how it does it picks out Buyer plans for a house • systems architecture: describes how a system works in terms of major subsystems. Contractor lines up Types of architectures plumbers, roofers… • technical architecture: describes how a system works in terms of implementation details, language Studs go here, pipes here… © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 8

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Focus of this Course • operational architecture: describes what the systems does, not how it does it • systems architecture: describes how a system works in terms of major subsystems • technical architecture: implementation details, language © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 9

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Often Changes with New Languages, Applications • operational architecture: describes what the systems does, not how it does it • systems architecture: describes how a system works in terms of major subsystems • technical architecture: implementation details, language Subject of Standards Wars © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 10

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Why Worry About Software Organization? • Recall at least 7 distinct areas of Artificial Intelligence, each with own algorithms and data structures; these have to be “knitted” together somehow • Software engineering is necessary for a successful software enterprise © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 11

Thinking About Architectures is Good Software Engineering 4 a • Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Abstraction – Ignores details to permit focus for thinking about general organization of intelligence – Semiformal description • Modularity – high cohesion (do one thing well, may be able to substitute different algorithms which produce same results “logical sensors”, “logical behaviors”, etc) – low coupling (which means may be able to add “apps”) – Supporting unit testing and debugging • Anticipation of change, Incrementality – How to adapt, support evolution • Generality – Not re-invent the wheel each time © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 12

4 a Operational Architecture of Biological Intelligence* Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary *An amazingly sweeping generalization for the purpose of metaphor © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 13

4 a Lower Central Nervous System Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Spinal Cord and “lower brain” Skills and responses *An amazingly sweeping generalization for the purpose of metaphor © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 14

4 a Loop 1: Reflexes, Reactive, Unconsciousness… Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Smart as a cockroach, fish, bird, armadillo… Spinal Cord and “lower brain” Skills and responses *An amazingly sweeping generalization for the purpose of metaphor © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 15

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Brain “Upper brain” or cortex Reasoning over symbols (information) about goals *An amazingly sweeping generalization for the purpose of metaphor © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 16

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Loop 2: Deliberative, Thoughtful, Conscious… “Upper brain” or cortex Reasoning over symbols (information) about goals Consciousness, Planning, optimization, situation awareness *An amazingly sweeping generalization for the purpose of metaphor © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 17

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Direct Perception to Symbols “Upper brain” or cortex Reasoning over symbols (information) about goals “Middle brain” Converting sensor data into symbols (information) Spinal Cord and “lower brain” Skills and responses *An amazingly sweeping generalization for the purpose of metaphor © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 18

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Two Loops Plus a Transform “Upper brain” or cortex Reasoning over symbols (information) about goals “Middle brain” Converting sensor data into information Spinal Cord and “lower brain” Skills and responses *An amazingly sweeping generalization for the purpose of metaphor © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 19

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary 2 Layers in Architecture “Upper brain” or cortex Reasoning over symbols (information) about goals “Middle brain” Converting sensor data into symbols (information) Deliberative Layer Reactive (or Behavioral) Layer Spinal Cord and “lower brain” Skills and responses © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 20

But There Is “Emotional Intelligence” Too 4 a • Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary User Interfaces – Displays, transparency of what robot is doing/thinking – Natural Language, gestures • Working in teams – Explicit multi-agent coordination often relies on “social rules” (though implicit swarm intelligence may not) • “Persona” we present to others – Security: what can you at your security grade can see about me – Human-robot interaction: affective responses, natural language *An amazingly sweeping generalization for the purpose of metaphor © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 21

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Canonical Architecture “Upper brain” or cortex Reasoning over symbols (information) about goals “Middle brain” Converting sensor data into symbols (information) Spinal Cord and “lower brain” Skills and responses Interaction Layer Deliberative Layer Reactive (or Behavioral) Layer © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 22

4 a Canonical Architecture Interaction: Persona and teaming Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Deliberative loop: Reasoning over symbols Key function: Converting sensor data into symbols Reactive loop: Skills and responses Interaction Layer Deliberative Layer Reactive (Behavioral) Layer © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 23

4 a What Does This Mean for Programming? • From Reactive to Deliberative Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary – Two types of perception: DIRECT, RECOGNITION (symbols) • Impacts computer vision – Different time horizons • From Present to Present, Past, Future • Impacts sensing, storage, as well as algorithms reasoning, projecting – Need a central structure (WORLD MODEL) to hold the symbols, history, knowledge but is tractable • From Reactive/Deliberative to Interaction – Additional knowledge “theory of mind”– beliefs, desires, intentions (BDI) of the other agent, common ground © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 24

4 a The Hard Parts • From Reactive to Deliberative Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary – Two types of perception: DIRECT, RECOGNITION (symbols) • Impacts computer vision – Different time horizons • From Present to Present, Past, Future • Impacts sensing, storage, as well as algorithms reasoning, projecting – Need a central structure (WORLD MODEL) to hold the symbols, history, knowledge but is tractable • From Reactive/Deliberative to Interaction – Additional knowledge “theory of mind”– beliefs, desires, intentions (BDI) of the other agent, common ground © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 25

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Can we make this more tangible? © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 26

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Consider as a Process: OODA • OODA loop, from military pilots – Observe, Orient, Decide, and Act – Problem: not everything intelligent requires “deciding” (ex. , reflexes or muscle memory such as riding a bike or driving a car) You Tube video on John Boyd and OODA http: //www. youtube. com/watch? v=iv. TBv 3 wnp 1 Yt © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 27

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Consider as a Process: AOP Can be Null • Action-Perception Cycle from cognitive science, artificial intelligence – Action-oriented perception is similar to OODA • stronger scientific basis and acceptance • “short cuts” for reflexes, reactive behaviors – AI has cast these as 4 components or primitives: SENSE, PLAN, ACT, and LEARN – Problem: too high level to be of much value in software organization; for example different complexities of what to sense (ex. proximity to an obstacle vs. obstacle is unusual for this situation and indicates an ambush) © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 28

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Primitives for Robot Intelligence SENSE PLAN ACT LEARN © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 29

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary “Behavioral Robotics” Reactive (or Behavioral) Layer sense act • “born” with sets of SENSE-ACT couplings called behaviors that get turned on/off based on stimulus • No PLAN © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 30

4 a Control Theory is “Lower Level” But Doesn’t Necessary Capture it All Reactive (fly by wire, inner loop control, behaviors): Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary • Tightly coupled with sensing, so very fast sense act • Many concurrent stimulusresponse behaviors, strung together with simple scripting with FSA • Action is generated by sensed or internal stimulus • No awareness, no mission monitoring • Models are of the vehicle, not the 31 “larger” world © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019)

4 a Behavioral Robotics • Focus in AI from 1986 -1996 Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary • Things you can do with behavioral robotics – Roomba, Aibo – Guarded motions (don’t hit anything), panic behaviors (stop!), “Macros” such as self-righting in teleoperation • Advantages – Direction perception is usually simple – High modularity, add new behaviors without reprogramming old working behaviors, degrades gracefully (depending on the technical architecture) © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 32

4 a Behavioral Robotics Issues • Hidden Costs Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary – Behaviors plus some sort of coordination function to fuse outputs to effectors • Where it breaks – “Fly at window” effects due to local scope (but lots of ways around this) – Poor choice of coordination functions (go left, go right = go in middle and hit obstacle) • Where it doesn’t break but scares/annoys people – Since sensor/situated based often can’t predict whether will go left of right to avoid (but only if doesn’t matter, see Na. Ts) – Not optimal because sensor based (reactive) versus symbol/map based © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 33

4 a “Hybrids” plan Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Deliberative Layer Reactive (or Behavioral) Layer World plan model sense act • PLAN, then instantiate appropriate SENSE-ACT behaviors, until next step in plan, … • PLAN requires a World Model (though it is bounded) plus the actual planning algorithms © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 34

4 a 3 Tier Variant plan Planner Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary World plan model Sequencer Reactive (or Behavioral) Layer sense act • PLAN, then instantiate appropriate SENSE-ACT behaviors, until next step in plan, … • Technical PLAN execution but a significant programming aspect © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 35

4 a LEARN • Different types of learning, different things to learn– is very diffuse so more permeates the architecture Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 36

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary So Far… • Biological organization suggests three layers of intelligence with distinctly different perception, knowledge, planning horizons, and time scales • The AI Robotics field has converged on PLAN, then SENSE-ACT with LEARN as needed at different points – Technically this is SENSE-PLAN, SENSE-ACT but historically the sensing for planning just like the execution monitoring is lumped in “PLAN” • But still not as tangible as desired… © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 37

4 a Consider as Functions • Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Large scale factory automation (not factory robots) has 4 functions – – • generating selecting implementing Monitoring May involve different contributions/combinations of primitives – Ex. monitoring requires sensing, understanding what was planned, and perhaps learning of what is normal and not normal • Factory automation functions don’t capture implementation but does seem to capture deliberation – what was implemented was usually a single, monolithic routine or control loop; natural intelligence usually have many low-level “routines” running concurrently • Problem: still missing data/knowledge structures; not prescriptive: if I want to do X, then pick these components or have this set of layers… © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 38

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary PLAN got Bigger and Matches 3 T monitoring generating World plan model selecting sense implementing act sense act • Upper level (PLANNER) • is mission generation & monitoring • Need Past, Present, Future • Lower level (SEQUENCING) • is selection of behaviors to accomplish task (instantiation) & local monitoring • Need Past, Present © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 39

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Other Ramifications: Planning Horizon, Time Scales monitoring generating World plan model selecting sense implementing act sense act PRESENT+PAST+FUTURE, SL PRESENT+PAST, FAST PRESENT, VERY FAST © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 40

Canonical Operational Architecture 4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Reactive Layer Interaction Layer Deliberative Layer • Each layer has a different style of program organization • Not concerned with details of interaction layer Behavioral Layer © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 41

4 a Programming Languages Reactive Layer Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Interaction Layer Procedural, Functional, Ontological languages such as OWL Deliberative Layer Functional languages such as Lisp Behavioral Layer Procedural languages such as C, C++, Java © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 42

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Advantages of programming in layers with different styles • Decomposition of a complex system – Can use a separate processor(s) for each layer or behaviors – Can split between on-board, off-board • Matching right tools and mindset for the task – Ex. C++ for behaviors, Lisp for planning • Add to working, verified code © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 43

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Given that autonomy has a different programming style, what is it? • There are three different styles as formalized by the 3 T hybrid operational architecture • The styles reflect 5 attributes: – Primitives involved (SENSE, PLAN, ACT) – Perception (direct or recognition) – Planning horizon (present, present+past+future) – Time scale (very fast, slow) – Models (local, global) © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 44

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Can intelligence be added in layers? Like upgrading to “pro version” or downloading “apps” as needed? © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 45

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Can intelligence be added in layers? Like upgrading to “pro version” or downloading “apps” as needed? • Yes, intelligence is organized in layers • Yes, Can create libraries of equivalent (“logical”) algorithms • But as will be seen in later lectures… – Adding new behaviors or algorithms is non-trivial • Coordination functions at reactive layer impose certain assumptions, restrictions, and side effects – Adding another layer is non-trivial • Different attributes such as perception, models require significant design investment • “Hidden” coordination between layers © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 46

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary HOW MUCH ARTIFICIAL INTELLIGENCE DOES A ROBOT NEED? © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 47

4 a How Much Artificial Intelligence Does a Robot Need? • It depends… Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary • What functions does the robot need to do? – Generate? Monitor? Select? Implement? Execute behaviors? Learn? • What planning horizon does the functions require? – Present, Present+Past+Future • How fast do the algorithms have to update? – May have to use a closed world and guaranteed execution rates (control theory) • What type of model does the robot need? Local? Global? Both? – Note: go with the minimum © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 48

4 a Note that “How Much” Rating Systems Aren’t Tied to Programming Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Am not picking on ALFUS, it’s just the most recent attempt © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 49

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary SUMMARY AND ADDITIONAL THOUGHTS © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 50

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary: Architectures • An architecture is the Big Picture of how to program an intelligent robot • “Architecture” refers to either – The operational architecture which described what the system does on an abstract level – The system architecture developed by a manufacturer (or research group) and may look like a data flow diagram – The technical architecture, specifies the actual techniques and code organization • The canonical operational architecture for an AI robot consists of three layers which generally represent different programming styles or even implementation languages: – Behavioral – Deliberative – Interaction © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 51

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Other Observations • Current practice is good with deliberative functions operating on symbols, good with behaviors using direct perception • Major barrier is going from sensory data to symbols: recognition and labeling as unique instances – May be able to see a coffee cup but not that it is my coffee cup and yours from the same dish collection is over there • Major barrier in understanding human intention (which is often implied but never spoken); AI robots currently require explicit directions © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 52

4 a Objectives Review Architecture -types Operational -Biological -Process -Functions -Ramifications Layered Int. How much? Summary Future Questions and Lectures • Since some aspects seem beyond current programming capabilities, what are levels of autonomy in terms of how much the robot can be trusted to do and how much the human has to do? • What is the systems architecture for the hybrid architecture? • Where do the 7 areas of AI fit in with the operational architecture? © 2019 Robin Murphy Introduction to AI Robotics 2 nd Edition (MIT Press 2019) 53