Toward a Design for Teaching Cognitive Robotics. Matthew D. Tothero Oskars J. Rieksts

Transcription

1 Toward a Design for Teaching Cognitive Robotics Matthew D. Tothero Oskars J. Rieksts

2 Criteria Embodied cognition Agent-principal paradigm Clear ontology Clear epistemology Concepts supporting agentprincipal interaction

3 Embodied Cognition Perceive Conceive Believe Achieve

4 Embodied Cognition Perceive receive & process sensa Conceive create concepts Gain understanding

5 Embodied Cognition Believe know or think to be true Achieve accomplish specific task

6 8 Laws of Embodied Cognition A robot cannot: conceive what it cannot perceive perceive what it cannot conceive achieve what it cannot conceive conceive what it cannot achieve

7 8 Laws of Embodied Cognition A robot cannot: conceive what it cannot believe believe what it cannot conceive perceive what it cannot believe believe what it cannot perceive

8 Agent-Principal Paradigm The robot is the agent The user is the principal The principal gives directives The robot carries out directives Interaction is required This requires communication

9 Metaphors of A-P Paradigm Hunter & hunting dog Note: continuous interaction General Eisenhower and President Roosevelt Directive: You will invade the European continent and defeat the Nazi war machine

10 Ontgy/Eptmy of A-P Paradigm Ontology & epistemology Hunter & hunting dog Overlap but not co-extensive Visual cortex vs. olfactory cortex Different conceptual structures Able to communicate with respect to task achievement

11 Ontgy/Eptmy of A-P Paradigm Eisenhower and Roosevelt Experientially disparate Different conceptual structures Different goals, but with overlap Able to communicate for task achievement

12 A-P Paradigm for Robotics Agent and principal are separate entities Agent acts on its own Agent receives and understands directives Agent and principal communicate

13 Implement A-P Paradigm Finch Raspberry Pi Laptop Client-server approach

14 Finch Carnegie Mellon's CREATE lab Parts: Light, temperature, and obstacle sensors Accelerometers Motors Multiple programming languages and environments

15 Wireless Finch Raspberry Pi USB HUB Battery Pack Wireless Adapter Kutztown Univeristy

16 Software Debian Python 3 OrientDB TCP/IP

17 Client/Server Approach Raspberry Pi - Server Laptop - Client Send commands to server Commands translated to Finch s API calls

18 GDB Software OrientDB Contained within Raspberry Pi

19 Finch Issues Not deterministic Straight line issues Video link

20 4tronix Diddyborg New Approaches

21 4tronix Initio

22 PiBorg Diddyborg

23 Hardware/Software Experiences Robots required initial setup 4tronix required the least amount of work, but the most amount of time Diddyborg required more technical skill such as soldering

24 Diddyborg Robot Comparison More durable and stronger Requires more sensors 4tronix Plastic Came with sensors out of the box

25 4tronix Video Link

26 Upcoming Tasks Additional sensors Indoor location services

27 Conclusion This approach shows promise Sensors and software must be tuned with respect to MMS theory (below) Drawbacks Cost per unit Not turnkey system

28 MMS Theory Affordances & constraints Drawbacks Cost per unit Not turnkey system

29 MMS Theory D/H A&C + software A&C determine mental model robot can construct Drawbacks Cost per unit Not turnkey system

30 MMS Theory Robot operates within (dynamic) mental model space

31 Bibliography, p. 1 Bickhard, Representational content in humans and machines, Journal of Experimental and Theoretical Artificial Intelligence, 5(4), 1993, R.A. Brooks, A robust layered control system for a mobile robot, IEEE Journal of Robotics and Automation, 2(1), 1986, TCP/IP R.A. Brooks, Intelligence without representation, Artificial Intelligence 47 (1991), M. Cowart, Embodied Cognition, Internet Encyclopedia of Philosophy, URL = <

32 Bibliography, p. 2 A. Kronfeld, Amichai. Reference and computation: an essay in applied philosophy of language (studies in natural language processing) (Cambridge, UK: Cambridge University Press, 1990). A. Noe. Action in Perception (Cambridge, MA: MIT Press, 2005). A. Noe. Spatial strategies in human-robot communication. Künstliche Intelligenz, 16(4), 2002,

33 Bibliography, p. 3 L. Shapiro. Embodied Cognition (New York, NY: Routledge, 2011). G. C. Smith. What Is Interaction Design? in B. Moggeridge. Designing Interactions (Cambridge, MA: MIT Press, 2007). G.M. Stratton, The spatial harmony of touch and sight, Mind, 8(32), 1899, R. A. Wilson and L. Foglia, Embodied Cognition, The Stanford Encyclopedia of Philosophy (Winter 2015 Edition), E.N. Zalta (ed.), URL = < died-cognition/>.