This presentation was given on the occasion of the defense of my M.Sc. thesis in computer science on February 1, 2006, at the Technical University of Munich (TUM).
Designing Libraries of Transparent, General and Restartable Activity Plans for a Simulated Autonomous Service Robot
1. Designing Libraries of Transparent, General
and Restartable Activity Plans for a Simulated
Autonomous Service Robot
Andreas J. Häusler
Intelligent Autonomous Systems Group
Institut für Informatik
Technische Universität München
February 1, 2006
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
2. Motivation Simulation Environment Plan Library Evaluation Summary
Outline
1 Motivation
2 Simulation Environment
3 Plan Library
4 Evaluation
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
3. Motivation Simulation Environment Plan Library Evaluation Summary
Outline
1 Motivation
2 Simulation Environment
3 Plan Library
4 Evaluation
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
4. Motivation Simulation Environment Plan Library Evaluation Summary
Robots Today
current stage
acting most accurately
highly specialized
do unpleasant, dangerous or often recurring work
drawbacks
focused on single tasks
hardware depends on application
acting in static environments
need supervision in most cases
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
5. Motivation Simulation Environment Plan Library Evaluation Summary
Robots Today
current stage
acting most accurately
highly specialized
do unpleasant, dangerous or often recurring work
drawbacks
focused on single tasks
hardware depends on application
acting in static environments
need supervision in most cases
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
6. Motivation Simulation Environment Plan Library Evaluation Summary
Challenges to Future Robots
robots of the future will have to be. . .
adaptable to dynamically changing environments
robust to failures
able to act on their own
able to accomplish a wide range of different tasks
much cheaper than today
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
7. Motivation Simulation Environment Plan Library Evaluation Summary
Challenges to Future Robots
robots of the future will have to be. . .
adaptable to dynamically changing environments
robust to failures
able to act on their own
able to accomplish a wide range of different tasks
much cheaper than today
task diversity
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
8. Motivation Simulation Environment Plan Library Evaluation Summary
Challenges to Future Robots
robots of the future will have to be. . .
adaptable to dynamically changing environments
robust to failures
able to act on their own
able to accomplish a wide range of different tasks
much cheaper than today
failure recovery
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
9. Motivation Simulation Environment Plan Library Evaluation Summary
Challenges to Future Robots
robots of the future will have to be. . .
adaptable to dynamically changing environments
robust to failures
able to act on their own
able to accomplish a wide range of different tasks
much cheaper than today
⇒ change from industrial to every-day household robots
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
10. Motivation Simulation Environment Plan Library Evaluation Summary
Requirements
The crucial point lies in how the robot’s behaviour is
implemented. It has to be. . .
transparent
general
restartable
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
11. Motivation Simulation Environment Plan Library Evaluation Summary
Requirements
The crucial point lies in how the robot’s behaviour is
implemented. It has to be. . .
transparent
general
restartable
transparency
inferences about actions
meaningful code structures
code abstraction
improved understandability
efficient learning
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
12. Motivation Simulation Environment Plan Library Evaluation Summary
Requirements
The crucial point lies in how the robot’s behaviour is
implemented. It has to be. . .
transparent
general
restartable
generality
adaptability
task reusability
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
13. Motivation Simulation Environment Plan Library Evaluation Summary
Requirements
The crucial point lies in how the robot’s behaviour is
implemented. It has to be. . .
transparent
general
restartable
restartability
failure recovery
task interruptability
repeatable results of task execution
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
14. Motivation Simulation Environment Plan Library Evaluation Summary
Outline
1 Motivation
2 Simulation Environment
3 Plan Library
4 Evaluation
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
15. Motivation Simulation Environment Plan Library Evaluation Summary
The Environment
dynamically changing
oriented at the real world
allows manipulation
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
16. Motivation Simulation Environment Plan Library Evaluation Summary
The B21 Robot
environmental sensors
internal sensors
actors
knowledge bases
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
17. Motivation Simulation Environment Plan Library Evaluation Summary
The B21 Robot
environmental sensors
camera
sonar sensors
laser sensor belt
internal sensors
actors
knowledge bases
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
18. Motivation Simulation Environment Plan Library Evaluation Summary
The B21 Robot
environmental sensors
internal sensors
touch sensors
rotation sensors
actors
knowledge bases
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
19. Motivation Simulation Environment Plan Library Evaluation Summary
The B21 Robot
environmental sensors
internal sensors
actors
robot motion actors
hand movement actors
arm movement actors
knowledge bases
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
20. Motivation Simulation Environment Plan Library Evaluation Summary
The B21 Robot
environmental sensors
internal sensors
actors
knowledge bases
environmental belief
entity-specific
knowledge
heuristics
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
21. Motivation Simulation Environment Plan Library Evaluation Summary
The B21 Robot
environmental sensors
internal sensors
actors
knowledge bases
required abilities
performing different
tasks
failure awareness Video
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
22. Motivation Simulation Environment Plan Library Evaluation Summary
Outline
1 Motivation
2 Simulation Environment
3 Plan Library
4 Evaluation
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
23. Motivation Simulation Environment Plan Library Evaluation Summary
The Concept Behind Activity Plans
Goals
Goal Routines
Low Level Routines
High Level Plans
Low Level Plans
Basic Control Commands
Lisp Processes
External Processes
composition
activity plans
goals
goal routines
low-level routines
collections of goal achievements
define what task has to be
fulfilled instead of how
allow the robot to know about
the task’s purpose
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
24. Motivation Simulation Environment Plan Library Evaluation Summary
The Concept Behind Activity Plans
Goals
Goal Routines
Low Level Routines
High Level Plans
Low Level Plans
Basic Control Commands
Lisp Processes
External Processes
composition
activity plans
goals
goal routines
low-level routines
directives for robotic actions
slots constitute “predicates” for
fulfillment tests
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
25. Motivation Simulation Environment Plan Library Evaluation Summary
The Concept Behind Activity Plans
Goals
Goal Routines
Low Level Routines
High Level Plans
Low Level Plans
Basic Control Commands
Lisp Processes
External Processes
composition
activity plans
goals
goal routines
low-level routines
specify actions for goal achieves
high-level plans contain failure
monitoring
low-level plans contain
“computations”
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
26. Motivation Simulation Environment Plan Library Evaluation Summary
The Concept Behind Activity Plans
Goals
Goal Routines
Low Level Routines
High Level Plans
Low Level Plans
Basic Control Commands
Lisp Processes
External Processes
composition
activity plans
goals
goal routines
low-level routines
directly controlling the robot’s
actors
contain no further goal achieves
act on the process level of the
robot’s architecture
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
27. Motivation Simulation Environment Plan Library Evaluation Summary
Designing the Plans
roadmap
1 thinking about the
requirements
2 introducing goal
inheritance
3 specifying the
execution order
4 construction
always keep in mind the
demanded behaviour
rembember to think about
possible failure situations
also consider recovery
approaches if applicable
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
28. Motivation Simulation Environment Plan Library Evaluation Summary
Designing the Plans
roadmap
1 thinking about the
requirements
2 introducing goal
inheritance
3 specifying the
execution order
4 construction
simplifies the implementation
process
enables “semantical”
structuring of the plans
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
29. Motivation Simulation Environment Plan Library Evaluation Summary
Designing the Plans
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
30. Motivation Simulation Environment Plan Library Evaluation Summary
Designing the Plans
roadmap
1 thinking about the
requirements
2 introducing goal
inheritance
3 specifying the
execution order
4 construction
allows bottom-up
implementation and testing
enables “semantical”
structuring of the control flow
allows further distinction
between high-level and
low-level plans
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
31. Motivation Simulation Environment Plan Library Evaluation Summary
Designing the Plans
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
32. Motivation Simulation Environment Plan Library Evaluation Summary
Designing the Plans
roadmap
1 thinking about the
requirements
2 introducing goal
inheritance
3 specifying the
execution order
4 construction
building a collection of
heuristical functions
always have in mind that
calls to heuristics and
trajectory computations
reside in low-level plans
failure monitoring and
recovery always are
contained in high-level plans
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
33. Motivation Simulation Environment Plan Library Evaluation Summary
Capability of the Plans
the demands
transparency
generality
restartability
others
achieving transparency through
proper implementation
splitting up the activity plans
externalizing the heuristical
functions
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
34. Motivation Simulation Environment Plan Library Evaluation Summary
Capability of the Plans
the demands
transparency
generality
restartability
others
thoughts about the achievement of
generality
coping with “new” situations
adapting to changed demands
splitting up the activity plans
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
35. Motivation Simulation Environment Plan Library Evaluation Summary
Capability of the Plans
the demands
transparency
generality
restartability
others
restartability through failure recovery
implementing sophisticated
failure monitoring
using the knowledge base
interruptable activity plans
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
36. Motivation Simulation Environment Plan Library Evaluation Summary
Capability of the Plans
the demands
transparency
generality
restartability
others
robustness of the plans
support for future learning
mechanisms
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
37. Motivation Simulation Environment Plan Library Evaluation Summary
Outline
1 Motivation
2 Simulation Environment
3 Plan Library
4 Evaluation
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
38. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
for ensuring high levels of generality and restartability
however, these cannot be tested separately
⇒ scenarios for the overall performance are needed
evaluation of transparency only through “reading the code”
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
39. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
for ensuring high levels of generality and restartability
however, these cannot be tested separately
⇒ scenarios for the overall performance are needed
evaluation of transparency only through “reading the code”
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
40. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
for ensuring high levels of generality and restartability
however, these cannot be tested separately
⇒ scenarios for the overall performance are needed
evaluation of transparency only through “reading the code”
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
41. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
for ensuring high levels of generality and restartability
however, these cannot be tested separately
⇒ scenarios for the overall performance are needed
evaluation of transparency only through “reading the code”
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
42. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
prerequisites to proper testing
main focus on failure monitoring and recovery
⇒ think about all failures the robot is aware of, and
the situations in which they will be invoked
how the robot has to react
low-level plan recovery only as consecutive failures
tests compare “normal” to “non-recovery” behaviour
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
43. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
prerequisites to proper testing
main focus on failure monitoring and recovery
⇒ think about all failures the robot is aware of, and
the situations in which they will be invoked
how the robot has to react
low-level plan recovery only as consecutive failures
tests compare “normal” to “non-recovery” behaviour
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
44. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
prerequisites to proper testing
main focus on failure monitoring and recovery
⇒ think about all failures the robot is aware of, and
the situations in which they will be invoked
how the robot has to react
low-level plan recovery only as consecutive failures
tests compare “normal” to “non-recovery” behaviour
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
45. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
prerequisites to proper testing
main focus on failure monitoring and recovery
⇒ think about all failures the robot is aware of, and
the situations in which they will be invoked
how the robot has to react
low-level plan recovery only as consecutive failures
tests compare “normal” to “non-recovery” behaviour
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
46. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
prerequisites to proper testing
main focus on failure monitoring and recovery
⇒ think about all failures the robot is aware of, and
the situations in which they will be invoked
how the robot has to react
low-level plan recovery only as consecutive failures
tests compare “normal” to “non-recovery” behaviour
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
47. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
the “pick-up”-scenario
for testing the stability of low-level plans
randomized robot- and object-positions
in each pass losing the object had been provoked
success if the robot managed to lift the object again
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
48. Motivation Simulation Environment Plan Library Evaluation Summary
Evaluation Scenarios
the “pick-up”-scenario
for testing the stability of low-level plans
randomized robot- and object-positions
in each pass losing the object had been provoked
success if the robot managed to lift the object again
the “move-entity”-scenario
testing the robot’s general performance
randomized starting positions
success if the object was placed on the table
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
49. Motivation Simulation Environment Plan Library Evaluation Summary
Results of the Evaluation
performance of pick-up-cup actions; entity-lost failure provoked
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
50. Motivation Simulation Environment Plan Library Evaluation Summary
Results of the Evaluation
performance of put-down-cup actions
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
51. Motivation Simulation Environment Plan Library Evaluation Summary
Results of the Evaluation
performance of the robot’s arms when picking up a cup
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
52. Motivation Simulation Environment Plan Library Evaluation Summary
Summary
reflecting the requirements
transparency through introducing a goal hierarchy
generality through heuristical functions
restartability through failure recovery
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
53. Motivation Simulation Environment Plan Library Evaluation Summary
Summary
reflecting the requirements
transparency through introducing a goal hierarchy
generality through heuristical functions
restartability through failure recovery
results of the work
universal approach for implementing activity plans
improved performance through failure recovery
appearance of (rudimentary) emergent behaviour
the library provides a reliable base for further work
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
54. Motivation Simulation Environment Plan Library Evaluation Summary
Summary
perspectives for the future
improving the present plans
adding new plans and new kinds of behaviour
inserting new technologies
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
55. Motivation Simulation Environment Plan Library Evaluation Summary
Summary
perspectives for the future
improving the present plans
adding new plans and new kinds of behaviour
inserting new technologies
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
56. Motivation Simulation Environment Plan Library Evaluation Summary
Summary
perspectives for the future
improving the present plans
adding new plans and new kinds of behaviour
inserting new technologies
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler
57. Motivation Simulation Environment Plan Library Evaluation Summary
Summary
perspectives for the future
improving the present plans
adding new plans and new kinds of behaviour
inserting new technologies
In our vision, the robot shall sometime be like a “human butler”
to private households.
Intelligent Autonomous Systems
Department of Informatics
Technische Universität München
Designing Activity Plan Libraries
Andreas J. Häusler