Midterm Presentation Spring 2005

1. Midterm Presentation Spring 2005 Tom Wilson Imaging, Robotics, & Intelligent Systems Laboratory University of Tennessee

2. Spring 2005 Midterm Presentation Outline Part I: General Overview • Progress – Spring 2005 • Research Overview • Tasks Description • Task 1 • Task 2 • Task 3&4 • Time-Management Profile • Semester Task Planning • Thesis Milestone Dates Part II – Focus on Current Research • Task 2: Robotic Architectures Part III – Summary • Conclusion and Questions

3. Part I – General Overview Progress - Spring 2005 • Completed:Task 1(100%) • On Schedule:Task 2 (90%) • On Schedule: Task 3-4 (beginning)

4. Part I – General Overview Research Overview All Spring 2005 research pertains to the development of a thesis on: Modular Robotic Systems (MRS)

5. (Sensors) (Platforms) Part I – General Overview Research Overview Modular Robotic Systems (MRS) (Systems)

6. Task 1 Part I – General Overview Selected Internal Reports • Format, Edit and Revise: Objective: Edit and revise (as needed) the relevant internal reports pertaining to the development of a thesis on the IRIS Modular Robotic System. • Purpose: This work parallels the Fall 2004 Task of editing & revising internally generated reports that pertain to the development of a thesis on a Modular Robotic System. • Fall 2004: • Internal reports (Systems) • Brick PILOT reports (Sensors)

7. Task 1 Part I – General Overview • Format, Edit and Revise: • Integration (Systems) • Radar (Sensors) • Mobility (Platforms) • Fusion (Systems) • Path Planning (Systems)

8. Part I – General Overview BeforeTask 1 - Development of Thesis

9. Part I – General Overview AfterTask 1 - Development of Thesis

10. Task 2 Part I – General Overview • Generate a ten (10) page report on: Robotic Architectures Objective: Develop a report using seminal texts in the field of Robotic Architecture. Purpose: Emphasize the use and integration of the paper as the first chapter of the thesis developed on the IRIS Modular Robotic System.

11. Part I – General Overview AfterTask 2 - Development of Thesis

12. Task 3&4 Part I – General Overview • Develop a thesis on: Modular Robotic Systems

13. Part I – General Overview Semester Task Planning

14. Part I – General Overview Task 3/4 - Thesis Milestone Dates

15. Part I – General Overview Task 3/4 - Thesis Milestone Dates

16. Part I – General Overview Task 3/4 - Thesis Milestone Dates

17. Part I – General Overview Task 3/4 - Thesis Milestone Dates

18. Part I – General Overview Task 3/4 - Thesis Milestone Dates

19. Part II – Current Research Task 2 - Generate a ten (10) page report on: Robotic Architectures (Chapter 1 MRS Thesis)

20. Part II – Current Research Task 2 - Outline 1. Introduction and Motivation 1.1 Introduction and Motivation 1.2 Introduction to AI Robotics Systems 1.3 Research Scope 1.4 Definitions and Terminology 1.5 The Organization of Intelligence: An Overview of AI Robotic Paradigms 1.6 Introduction to Architectures for AI Robotics 2. Literature Survey 3. Overview of Intelligent Control 3.1 Main Concepts 3.2 A general introduction to control systems that incorporate intelligent methods to attain higher degree of autonomy 4. An Overview of Intelligent Control Architectures 4.1 Hierarchical and Model-Based Architectures 4.2 Distributed Architectures 4.3 Knowledge-Based Architectures 4.4 Reactive Architectures 4.5 Subsumptive Architectures 4.6 Hybrid Architectures 5. A General Approach to the Design and Control of Intelligent Systems 5.1 An intelligent system design based on Real-Time Control System (Albus) 5.2 A model-based architecture for the design and simulation of high autonomy systems 6. Application to Modular Robotics Systems

21. Part II – Current Research Task 2 - % completed 1. Introduction and Motivation 1.1 Introduction and Motivation 1.2 Introduction to AI Robotics Systems 1.3 Research Scope 1.4 Definitions and Terminology 1.5 The Organization of Intelligence: An Overview of AI Robotic Paradigms 1.6 Introduction to Architectures for AI Robotics 2. Literature Survey 3. Overview of Intelligent Control 3.1 Main Concepts 3.2 A general introduction to control systems that incorporate intelligent methods to attain higher degree of autonomy 4. An Overview of Intelligent Control Architectures 4.1 Hierarchical and Model-Based Architectures 4.2 Distributed Architectures 4.3 Knowledge-Based Architectures 4.4 Reactive Architectures 4.5 Subsumptive Architectures 4.6 Hybrid Architectures 5. A General Approach to the Design and Control of Intelligent Systems 5.1 An intelligent system design based on Real-Time Control System (Albus) 5.2 A model-based architecture for the design and simulation of high autonomy systems 6. Application to Modular Robotics Systems

22. Part II – Current Research AfterTask 2 - Development of Thesis

23. Part II – Current Research Task 2 • My direct contribution to the MRS Thesis. • Serves as the “binding” element to the MRS Thesis • Intends to answer the question: Why Modular Robotics? => Problem Statement for MRS Thesis. • Should validate (or invalidate) the modular approach implemented in the IRIS Laboratory Research. • Provide insight to the strengths and weaknesses of MRS. • Finished Report:~40-50 pages (10 page assignment).

24. Part II – Current Research Task 2 – Current Work • 1. Introduction – What is this research about? • The aim of this report is to introduce and explain the theories and fundamental aspects of system architecturesforartificial intelligence (AI) robotic systems. • AI robotic systems recognize the environment and can execute tasks that are more dexterous, in environments that are more complicated, than non-intelligent systems. • Intelligent systems with high degrees of autonomy should perform well under significant uncertainties in the system and environment for extended periods, and should be able to compensate for certain system failures without external intervention. • Intelligent control systems evolve from conventional control systems by adding intelligent components.

25. Part II – Current Research Task 2 – Current Work • 2. Motivation – Why is this research being done? • This material serves as the binding element for the MRS Thesis. • Intends to answer the question: Why Modular Robotics? => Problem Statement for MRS Thesis. • Validates (or invalidates) the modular approach implemented in the IRIS Laboratory Research. • Provide insight to the strengths and weaknesses of MRS.

26. Part II – Current Research Task 2 – Current Work • 3. Literature Survey • Material Direction: seminal textbooks in AI Robotic Architecture (total of 7 texts). • Philosophical Overview • Albus - rejects behaviorist philosophy; seeks the creation of a mechanism that resembles the structure of the human mind (massive parallelism). Coined: “Outside-In.” • Arkin – embraces behaviorist viewpoint. Advocates of solutions from the “Inside-Out.” • May be others

27. Part II – Current Research Task 2 – Current Work • 3. Literature Survey (con’t.) • [1] Intelligent robotic systems: design, planning, and control, Jacak, Witold. Kluwer Academic / Plenum Publishers, New York (1999). • [2] Introduction to AI Robotic Systems, Murphy, Robin R. MIT Press, Cambridge, Massachusetts (2000). • [3] Behavior-Based Robots, Arkin, R. MIT Press, Cambridge, Massachusetts (1998) • [4] A Reference Model Architecture for Intelligent Systems Design, Albus, James S. Taken from: An Introduction to Intelligent and Autonomous Control, Antsaklis, Panos J. and Kevin M. Passino eds. Kluwer Academic Publishers, Norwell, Mass. (1993) • [5] Engineering of mind: an introduction to the science of intelligent systems, Albus, James S. and Alexander M. Meystel. John Wiley and Sons, NY, NY (2001). • [6] Robotic Engineering: an Integrated Approach, Klafter, Richard D, Thomas A. Chmielewski, Michael Nevgin. Prentice-Hall, Inc., Englewood Cliffs, New Jersey (1989). • [7] Minimalist Mobile Robotics: A Colony-style Architecture for an Artificial Creature. Connell, Jonathon H. Academic Press, Inc., San Diego, CA (1990).

28. Part II – Current Research Task 2 – Current Work • 3. Literature Survey (con’t.) • [1] Intelligent robotic systems: design, planning, and control, Jacak, Witold. Kluwer Academic / Plenum Publishers, New York (1999). • [2] Introduction to AI Robotic Systems, Murphy, Robin R. MIT Press, Cambridge, Massachusetts (2000). • [3] Behavior-Based Robots, Arkin, R. MIT Press, Cambridge, Massachusetts (1998) • [4] A Reference Model Architecture for Intelligent Systems Design, Albus, James S. Taken from: An Introduction to Intelligent and Autonomous Control, Antsaklis, Panos J. and Kevin M. Passino eds. Kluwer Academic Publishers, Norwell, Mass. (1993) • [5] Engineering of mind: an introduction to the science of intelligent systems, Albus, James S. and Alexander M. Meystel. John Wiley and Sons, NY, NY (2001). • [6] Robotic Engineering: an Integrated Approach, Klafter, Richard D, Thomas A. Chmielewski, Michael Nevgin. Prentice-Hall, Inc., Englewood Cliffs, New Jersey (1989). • [7] Minimalist Mobile Robotics: A Colony-style Architecture for an Artificial Creature. Connell, Jonathon H. Academic Press, Inc., San Diego, CA (1990).

29. Part II – Current Research Task 2 – Current Work • 4. Scope of Research – What is an intelligent robot? • Artificial Intelligent (AI) Robotics form a distinct field both historically and in scope from other intelligent systems such as Industrial Robots. • Research issues in the field of industrial robotics are concentrated on control theory issues, particularly solving the dynamics and kinematics of a mobile robotic appendage with a stationary base. • Research issues in the field of AI Robotics has concentrated on how a mobile robot should handle unpredictable events in an unstructured world.

30. Part II – Current Research Task 2 – Current Work • 4. Scope of Research (cont.) – Definition of intelligent robot • Murphy attributes AI as being “…the attempt to get the computer to do things that, for the moment, people are better at.” [2, p.15]. • Jacak describes a robotic systems and its control as intelligent if the system “…can self-determine its decision choices based upon the simulation of needed solutions in virtual world or upon experience gained in the past both from failures and successful solutions which are stored in the form of rules in the system knowledge base. • Albus uses an extensive, multipart definition and integrates the concept of intelligent control to describe an intelligent system…

31. Part II – Current Research Task 2 – Current Work • 4. Scope of Research (cont.) – Definition of intelligent robot • Albus’ definition: using the concept of intelligent control to describe an intelligent system… • Highly intelligent behavior is the result of goals and plansinteracting at many hierarchical levels with knowledge represented in a multi-resolution world model. • High levels of intelligence require a rich dynamic world model that includes both a priori knowledge and information provided by sensors and a sensory processing system. • Intelligent decision making requires a value judgment system that can evaluate what is good and bad, important and trivial, and can estimate cost, benefit, and risk of potential future actions • A model of intelligence can enable the engineering design of intelligent systems that pursue goals, imagine the future, make plan, and react to what they see, fell, hear, smell, and taste.

32. Part II – Current Research Task 2 – Current Work • 4. Scope of Research (cont.) – Definition of intelligent robot • Albus’ definition: • Summation: These actions will enable the development of systems that behave as if they are sentient, knowing, caring, creative individuals motivated by hope, fear, pleasure, pain, love, hate, curiosity, and a sense of priority.

33. Part II – Current Research Task 2 – Current Work • 4. Scope of Research (cont.) – Definition of intelligent robot • Other Commonly Used Terms and Definitions in AI • Turing Test: A machine is intelligent if it can communicate with a human without the human being able to differentiate the communication from a human (either via teletype, or otherwise). • ELIZA project (M.I.T.): Uncovered fallacy with the Turing Test – a algorithmic program was derived that marginally passed for human communication, demonstrating human-like intelligence is easily faked. • AI entity: What would you (the human) like to talk about? • Human: Let’s talk about my family. • AI entity: How is your family? Do you have any brothers or sisters. Etc…

34. Part II – Current Research Task 2 – Current Work • 4. Scope of Research (cont.) – Design Issues • Broadly, the design of an AI Robotics system considers: • How the robot will represent knowledge about the world. How to represent knowledge that can be used instead of merely being reproduced. • Whether the system needs to understand natural language. The difficulty is getting AI entities (computers) to understand and process human languages, or the ability to translate from one human language to another. • Implementing artificial vision: the difficulty lies in getting the system to recognize the objects it ‘sees.’ • Whether the system can learn tasks. • What kind of planning and problem solving will the system have to do. • How much inference is expected. • How rapidly can the system search its database and knowledge for answers • What mechanisms (sensors) the system will use for perceiving the world.

35. Part II – Current Research Task 2 – Current Work • 4. Scope of Research (cont.) – Contemporary Issues • At the present, the two types of intelligent robotic systems appear to be merging: • Industrial robotics is currently shifting to a “mass customization” phase where companies make short runs of special order goods. • The emphasis on these types of systems is to be able to be rapidly reprogrammed and more forgiving if a part is not placed exactly as expected in its workspace. • As a result of the points above, AI techniques are migrating to industrial robotics.

36. Part II – Current Research Task 2 – Current Work Planetary Rovers (highly autonomous) AI Robotics Vision Tele-systems Telemanipulators Industrial / Manufacturing Industrial / manipulators 1960 1970 1980 1990 2000

37. Part II – Current Research Task 2 – Current Work • 4. Scope of Research (cont.) – A Brief History • Industrial Manipulators were the first form of robot. • Teleoperation arose as an intermediate solution to tasks that required automation but for which robots could not be adequately programmed to handle. • Telepresence techniques attempt to create a more natural interface for the human to control the robot and interpret what it is doing and seeing, but at a high communication cost. • Supervisory control attempts to delegate portion of the task to the remote, either to do autonomously (traded control ) or with reduced, but continuous, human interaction (shared control).

38. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms • What is a robotic paradigm – an Overview? • AParadigm is a method of organizing the cognitive and physical actions of a system (an intelligent entity). • ThePrimitives: SENSE,PLAN,ACT. • A robotic paradigm is defined bythe relationship between the three primitives(SENSE, PLAN, ACT) andthe way sensory data is processed and distributedthrough the system.

39. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Fundamental Issues • AnArchitecture provides a set of principles for organizing a control system. • Must address Task and Domain Specificity– how suitable is it to a broad range of applications? • Example: an architecture well suited for direct teleoperation tends to not to be easily controlled for supervisory control or autonomous use.

40. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Major Classifications • Hierarchical (Deliberative) - look-ahead: sense, think/plan, then act. • Reactive - no look-ahead: react only. • Behavior-based - distribute thinking over acting. • Hybrid - combine 1+2, think slowly, react quickly. • Pure, hierarchical (deliberative) control is no longer used!

41. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Notes on Design Issues • It is impossible to tell what architecture is being used in any sufficiently complicated robot. • In systems with complex behaviors, the control architecture becomes very important. • The different properties of an environmentthat will impact the robot's control and therefore the choice of controlarchitecture. • Similarly, the properties of the robot's task impact the choice of the control architecture. The task requirements can constrain the architecture choice.

42. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Criteria for the selection of a Control Architecture • Support for parallelism – the ability of the architecture to execute processes/behaviors at the same time • Hardware targetability – how well the architecture can be mapped onto ‘real’ robot sensors and, how well the computation can be mapped onto ‘real’ processing elements. • Run-time flexibility – does the architecture allow run-time adjustment and reconfiguration? It is important for adaptation/learning. • Modularity – how does the architecture address encapsulation of control, how does it treat abstraction? Does it allow many levels, going from feedback loops to primitive to agents? Does it allow re-use of software?

43. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Criteria for the selection of a Control Architecture (cont’t.) • Niche Targetability – how well the architecture allows the robot to deal with its environment. • Robustness – how well does the architecture perform if individual components fail? How well does it enable and facilitate writing controllers capable of fault tolerance? • Ease-of-use – how easy to use and accessible is the architecture? Are there programming tools and expertise? • Performance – how well does the robot perform using the architecture? Does it act in real-time? Does it get the job done? Is it failure-prone?

44. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Summary and Conclusion: Criteria for the selection of a Control Architecture • The above issues allow the comparison and evaluation of different architectures relative to specific robotic designs, tasks, and environments. • However, not all tasks, environments, and designs are comparable.

45. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Distinguishing Control Features – Time Scale • Reactive systemsrespond to the real-time requirements of the environment. • Deliberative systemslook ahead (plan), and thus work on a longer time-scale. • Hybrid systemsmust combine the two time-scales in an effective way, usually requiring a middle layer; consequently they are often called three-layer architectures. • Behavior-based systemsattempt to bring the different time-scales closer together by distributing slower computation over concurrent behavior modules. • Time-scaleis an important way of distinguishing control architectures.

46. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Distinguishing Control Features – Representation • Another key distinguishing feature between architectures is representation of the world/environment, also called world modeling. • Some tasks and architectures involve storing information about the environment internally, in the form of an internal representationof the environment. • For example, while exploring a maze, a robot may want to remember a sequence of moves it has made (e.g., "left, left, right, straight, right, left"), so it can back-track and find its way. Thus, the robot is constructing a representation of its path through the maze. • The robot can also build a mapof the maze, by drawing it using exact lengths of corridors and distances between walls, etc. This is also a representation of its environment, a model of the world. • If two robots are working together, and one is much slower than the other, if the fast robot remembers/learns that the other is always slower, that is also a type of a model of the world, in this case, a model of the other robot.

47. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Distinguishing Control Features – Different representations of World Models • There are numerous aspects of the world that a robot can represent/model, and numerous ways in which it can do it, including: • Spatial,metric or topological: maps, navigable spaces, structures. • Objects: instances of detectable things in the world. • Actions:outcomes of specific actions on the self and environment. • Self/ego:stored proprioception: sensing internal state, self- limitations, etc. • Intentionalgoals, intended actions, plans. • Symbolicabstraction: encoding of state/information.

48. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Distinguishing Control Features – Amount and Type of Representation • The amount and type of representation or modeling used by a robot is critically related to the type of control architectureit is using. • Some models are very elaborate; they take a long time to construct and are therefore they are kept around possibly throughout the lifetime of the robot's task (for example, a task that makes detailed metric maps). • Others may be relatively quickly constructed and be used in a transient capacity. These are used quickly and discarded or updated (a short plan, or perhaps an immediate goal, etc.). • How long it takes to construct/build a model is an important aspect of the robot's controller.

49. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Distinguishing Control Features – Amount and Type of Representation (con’t.) • How long it takes to use it (the representation) is an equally important consideration. • Consider making detailed metric maps again: • It takes a long time to construct an accurate and detailed metric map, because it requires exploring and measuring the environment. • Furthermore, it takes time to use such a map (even if it took no time to construct it, and it was given to the robot by the designer). • One must find the free/navigable spaces in the map, and then search through those to find the best path to the goal. • Similarly, any internal model can require time to construct and be used, and these timing requirements directly affect the time-scale of the controller.

50. Part II – Current Research Task 2 – Current Work • 5. The Organization of Intelligence: An Overview of Robotic Paradigms (con’t.) • Control Architectures and Internal States/Representations • A control architecture can make it easy or difficult to store and manipulate internal world models (just as a programming language can make it more or less convenient to build and store structures), i.e., compute with them. • How an internal state (i.e., the information a robotic system retains) relates to representation. • In principle, any internal state is a form of representation. • What matters is the form and function of that representation. • Staterefers to the "status" of the system itself • Whereas "representation" refers to arbitrary informationthat may be contained in the system.