Incremental Causal Connection For Self-Adaptive Systems Based On Relational Reference Attribute Grammars

René Schöne, Johannes Mey, Sebastian Ebert, Sebastian Götz, Uwe Aßmann
Incremental Causal Connection for Self-Adaptive

Systems Based on Relational Reference Attribute
Grammars
Montréal, October 26th 2022
A Case Study
Site A Site B
R1
2
R1 R2
4
1 R3
3
Collabo-
R2 ration
1 B1 Robot Zone
Operator
Controller
Scene
Controller A R3
2 3 4 B2 Robot
Controller
MovableObject Region Component Causal Connection Port Port Target
Incremental Causal Connection for Self-Adaptive Systems Based on Relational Reference Attribute Grammars
René Schöne, Johannes Mey, Sebastian Ebert, Sebastian Götz, Uwe Aßmann Slide 2 of 14
A Case Study
Site A Site B
R1
2
R1 R2
4
1 R3
3
Collabo-
R2 ration
1 B1 Robot Zone
Operator
Controller
Scene
Controller A R3
2 3 4 B2 Robot
Controller
scenecommand
command
scene
scene
RAG Model A logical scene RAG Model B
Combining Two Concepts
Self-Adaptive Systems (SAS) Reference Attribute Grammars (RAG) [Hed00]
• Modern software: Start small. Gets
complex, distributed, heterogeneous • Based on Attribute Grammars [Knu68]
• Goals: Separation of concerns, fast • Origin: compiler construction (program
prototyping, theoretical grounding, analysis)
efficiency • Advantages: concise specification, efficient
• One approach: Model-driven evaluation, natural separation of concerns
engineering / Models@run.time
What if: we use RAGs as technological space for the model?
Challenges Ahead
Specifications of Managing System
• modular
Self-adaptive System
• (re-)usable
read
Adaptation Goals
Managing
System
• understandable
• correct
adapt
sense • efficient
Managed System
Connections from Managing System

• modular
• declarative
sense effect
sense
Environment • reusable
[Wey19]
Challenges Ahead
Self-adaptive System Specifications of Managing System

• modular
read
Adaptation Goals
• (re-)usable
Managing
System
• understandable
• correct
adapt
• efficient
sense
Managed System

sense effect
• modular
• declarative
sense
Environment
• reusable
[Wey19]
Challenges Ahead
Specifications of Managing System
• modular
Self-adaptive System
• (re-)usable
read
Adaptation Goals
Managing
System
• understandable
• correct
adapt
sense • efficient
Managed System

• modular
• declarative
sense effect
sense
Environment • reusable
[Wey19]
Model Specification using Attribute Grammars (AGs)
Attribute Grammars specify semantics/behaviour for a tree of a context-free grammar.
We use the JastAdd object-oriented Reference AG system [HM03].
• Efficient evaluation
Tree Structure Declarative Specification
• Concise specification along tree
inherited (↓), synthesized (↑) attributes – Incremental evaluation
– Concurrent evaluation
• Collections (JastAdd)
• Circular attributes, AOP features (JastAdd)
Grammars as Models
• In models:
Main Difference: Relations RC
– Containment relations form overlay tree

L L
– Non-containment relations
Rou. Reg.
– Bidirectional relations
• In grammars:
L L
}
L
– Containment references:
SwP. Sen.
– Non-containment references
Seg.
AST
– Bidirectional references
Relational RAGs [Mey+20] Seg. Sw. Seg.
Sem. Sem.
Grammars as Models
WorldModel ::= [MyScene:Scene]

WorldModel
ExecutedOperation:Operation*
/NextOperation:Operation/;
NextOperation MyScene
Position ::= <X:double> <Y:double> <Z:double>; ExecutedOperation
1 0 .. 1
* Object
Scene ::= DropOffLocation* MovableObject*; Operation Scene
String Name
abstract Object ::= <Name:String> Position;

1
DropOffLocation : Object ::= ; Position
MovableObject : Object ::= ; * * double X
double Y
abstract Operation ; PickAndPlace DropOffLocation MovableObject double Z
1 TargetLocation
PickAndPlace : Operation ::= ; 1 ObjectToPick
rel PickAndPlace.ObjectToPick -> MovableObject;

rel PickAndPlace.TargetLocation -> DropOffLocation;
Grammars as Models

WorldModel
1 0 .. 1
* Object
String Name

1
double Y
1 TargetLocation

Grammars as Models

WorldModel
1 0 .. 1
* Object
String Name

1
double Y
1 TargetLocation

Grammars as Models

WorldModel
1 0 .. 1
* Object
String Name

1
double Y
1 TargetLocation

Grammars as Models

WorldModel
1 0 .. 1
* Object
String Name

1
double Y
1 TargetLocation

Grammars as Models

WorldModel
1 0 .. 1
* Object
String Name

1
double Y
1 TargetLocation

Grammars as Models

WorldModel
1 0 .. 1
* Object
String Name

1
double Y
1 TargetLocation

Grammars as Models

WorldModel
1 0 .. 1
* Object
String Name

1
double Y
1 TargetLocation

An Iterative Process to Build Self-Adaptive Systems
add/refine computations Structure

(Grammar)
revisit structure
Computation
(Attributes)
An Iterative Process to Build Self-Adaptive Systems
add/refine computations Structure

(Grammar) revisit structure
revisit structure
Computation add/refine Semantics

(Attributes) semantics (Connections)
Coming Back to our Case Study
Site A Site B
R1
2
R1 R2
4
1 R3
3
Collabo-
R2 ration
Operator
1 B1 Robot Zone
Controller
Scene
Controller A R3
2 3 4 B2 Robot
Controller
command
scene
command
scene
scene
Coming Back to our Case Study
Site A Site B
R1
2
R1 R2
4
1 R3
3
Collabo-
R2 ration
Operator
1 B1 Robot Zone
Controller
Scene
Controller A R3
2 3 4 B2 Robot
Controller
command
scene
command
scene
scene
ect
Ports Mark Required Connections for our Case Study
AST as model@run.time
send port Region
WorldModelB
OtherScene [0] LogicalScene
name = "A"
receive port
② NextOperation MyScene
Robot
OwnedCollaborationZoneNames
OccupiedCollaborationZoneNames
RobotToExecute
PickAndPlace ① Scene LogicalMovableObject LogicalRegion
CurrentPosition
Reachability [0] ObjectToPick
CanReachObjectOfInterest MovableObject LogicalScene

DropOffLocation
ObjectName = "blue2" Name = "blue2"
Driver Code
ect
Ports Mark Required Connections for our Case Study
AST as model@run.time
send port Region
WorldModelB
OtherScene [0] LogicalScene
name = "A"
receive port
Robot
RobotToExecute
CurrentPosition
CanReachObjectOfInterest MovableObject LogicalScene

DropOffLocation
Driver Code
send port
A DSL to Specify Connections
Region
WorldModelB
OtherSce
name = "A"
receive port
send WorldModelB.NextOperation using PrintOperation ;
PrintOperation maps Operation op to byte[] {: ② NextOperation MyScen
byte[] result = op.toProtobufByteArray();
if (result == null) { reject(); }
Robot
①
return result;
OwnedCollaborationZoneNames RobotToExecute
:} PickAndPlace Scene
CurrentPosition
Port target Recompute Apply Different yes Send
has changed port target mappings value?
message
Recognized by incremental if rejected no

evaluation feature of JastAdd Message
CanReachObjectOfInterest MovableObject
DropOffLoc
Supported Grammar Elements of the DSL
Support for Different Parts of the Grammar Using Send
and Receive Ports. Legend: Full support, G
# Partly
supported, # Not supported, / not applicable
Element of a grammar Send Receive
Terminal
Non-Terminal (Context-free)
Non-Terminal (Single)
Non-Terminal (List)
Non-Terminal (Optional)
Relation G
# #
Attribute /
✅ Declarative and concise textual specification
olves manual work
⬋ ✅ Similar to a metamodel using [Mey+20]
Detailed Generation Process
Compu- Driver
Grammar
❌ Notation Code
uses [Sch+20])
connections to other systems possi
cation ⬋
⬊ (only for simple tokens
uation
Connect Connection Generated
Specification
RagConnect
Aspect JastAdd
Code
HM03] + Relational RAGs [Mey+20] + RagConnect
1st phase: Specification 2nd phase: Generation 3rd phase: Usage
Grammar (Structure) AST as model@run.time
send port Region OtherScene [0] L
.
① receive port
name = "A"
WorldModelB
②
Robot
RobotToExecute
PickAndPlace ① Scene Logic
CurrentPosition
René Schöne, Johannes Mey, Sebastian Ebert, Sebastian Götz, Uwe Aßmann Reachability [0] Slide 12ObjectToPick
of 14
olves manual work
Compu- Driver
Grammar
❌ Notation Code
uses [Sch+20])
cation ⬋
uation
Specification
RagConnect
Aspect JastAdd
Code
WorldModelB ::= Region* Robot* ExecutedOperation:Operation
[MyScene:Scene] OtherScene:LogicalScene*
HM03] + Relational/NextOperation:Operation/
RAGs [Mey+20] + RagConnect ;
1st phase: Specification
abstract Operation ; 2nd phase: Generation 3rd phase: Usage
Grammar (Structure)
PickAndPlace : Operation ; AST as model@run.time
.
① receive port
name = "A"
WorldModelB
②
Robot
RobotToExecute
CurrentPosition
of 14
olves manual work
Compu- Driver
Grammar
❌ Notation Code
uses [Sch+20])
cation ⬋
uation
Specification
RagConnect
Aspect JastAdd
Code
WorldModelB ::= Region* Robot* ExecutedOperation:Operation
[MyScene:Scene] OtherScene:LogicalScene*
HM03] + Relational/NextOperation:Operation/
RAGs [Mey+20] + RagConnect ;
1st phase: Specification
abstract Operation ; 2nd phase: Generation 3rd phase: Usage
Grammar (Structure)
PickAndPlace : Operation ; AST as model@run.time
.
① receive port
name = "A"
WorldModelB
②
Robot
RobotToExecute
CurrentPosition
of 14
olves manual work
Grammar
✓ Compu- Driver
❌ Notation Code
uses [Sch+20])
cation ⬋
uation
Specification
RagConnect
Aspect JastAdd
Code
.
① receive port
name = "A"
WorldModelB
②
Robot
RobotToExecute
CurrentPosition
of 14
olves manual work
Grammar
✓
syn Operation WorldModelB.getNextOperation() {
Compu- Driver
❌ Notation Code
if (diffToOperations().getNumChild() == 0)
uses
return errorNoOperationComputed(); connections to other systems possi
cation ⬋
⬊ for (Operation op : diffToOperations()) { (only for simple tokens [Sch+20])
uation
JastAdd
Robot executingRobot = op.getRobotToExecute();
RagConnect
Specification Aspect Code
if (!op.isErrorOperation()) {
if (op.equals(lastOperationFor(executingRobot))) {
return errorDuplicateOperation();
HM03] + Relational} RAGs [Mey+20] + RagConnect
1st}phase: Specification 2nd phase: Generation 3rd phase: Usage
return op;
}
return errorNoExecutableOperation(); send port Region OtherScene [0] L
.
① } receive port
name = "A"
WorldModelB
②
Robot
RobotToExecute
CurrentPosition
of 14
olves manual work
Grammar
✓ Compu- ✓ Driver
❌ Notation Code
uses [Sch+20])
cation ⬋
uation
Specification
RagConnect
Aspect JastAdd
Code
.
① receive port
name = "A"
WorldModelB
②
Robot
RobotToExecute
CurrentPosition
of 14
olves manual work
Grammar
❌ Notation Code
uses;
cation ⬋
⬊ (only for simple tokens [Sch+20])
uation
Connect PrintOperation Connection Generated
send WorldModelB.NextOperation using PrintOperation
Specification
RagConnect
byte[] result =Aspect JastAdd
Code
maps Operation op to byte[] {:
op.toProtobufByteArray();
if (result == null) { reject(); }
HM03] + Relational RAGs [Mey+20]return+ RagConnect
result;
:}
.
① receive port
name = "A"
WorldModelB
②
Robot
RobotToExecute
CurrentPosition
of 14
olves manual work
Grammar
❌ Notation Code
uses [Sch+20])
cation ⬋
uation
Connect ✓ RagConnect
Connection
JastAdd Generated
.
① receive port
name = "A"
WorldModelB
②
Robot
RobotToExecute
CurrentPosition
of 14
olves manual work
Grammar
✓
WorldModelB model = new WorldModelB(); Compu- ✓ Driver
❌ Notation Code
model.ragconnectResetEvaluationCounter();
uses [Sch+20])
cation model.addOtherScene(new LogicalScene()); ⬋
uation
Connection
for (String topic : config.forB.topicsSceneUpdate)
JastAdd
{
Generated
Specification
model.connectMyScene(topic); Aspect Code
}
1stmodel.connectOtherScene("place-a/logical/update",
phase: Specification 2nd phase: Generation 0); 3rd phase: Usage
Grammar model.connectNextOperation("robot/operation",
(Structure) AST as model@run.time
false);
.
① receive port
name = "A"
WorldModelB
②
Robot
RobotToExecute
CurrentPosition
of 14
olves manual work
Grammar
✓ Compu- ✓ Driver ✓
❌ Notation Code
uses [Sch+20])
cation ⬋
uation
Connection
JastAdd Generated
.
① receive port
name = "A"
WorldModelB
②
Robot
RobotToExecute
CurrentPosition
of 14
Evaluation of Concisness
LOC of all Case Studies
es
ar
ut
t
m
ec
ib
m
va
r
on
ra
tt
Ja
Case Study Part
C
Common (manual) 15 255 52 318
Site A (manual) 1 17 2 136
Site A (generated) 31 2570 11 042
Main
Site B (manual) 29 668 136 350
Site B (generated) 69 5443 23 075
Main (manual) 15 261 8 144
Coordinator
Main (generated) 28 1510 9592
Evaluation of Concisness
LOC of all Case Studies
es
ar
ut
t
m
ec
ib
m
va
r
on
ra
tt
Ja
Case Study Part
C
Common (manual) 15 255 52 318
Site A (manual) 1 17 2 136
Site A (generated) 31 2570 11 042
Main
Site B (manual) 29 668 136 350
Site B (generated) 69 5443 23 075
Main (manual) 15 261 8 144
Coordinator
Conclusion
This work presents

• a way to connect RAG-based models with external systems
• easily, with a DSL
• efficiently, exploiting RAG technology
We showed this
• in a robot coordination case study
• . . . for solving a cooperative task
• . . . and for coordinating the runtime components.
We provide the missing concepts to use RAGs for Models@run.time!
Visit us at our poster!
Conclusion
Incremental Causal Connection for Self-Adaptive Systems
Based on Relational Reference Attribute Grammars
This work presents

• a way to connect RAG-based models with external systems
René Schöne* (✉ rene.schoene@tu-dresden.de) // Johannes Mey* (✉ johannes.mey@tu-dresden.de) //
Sebastian Ebert*,‡ (✉ sebastian.ebert@tu-dresden.de) // Sebastian Götz* (✉ sebastian.goetz1@tu-dresden.de) //
Uwe Aßmann*,‡ (✉ uwe.assmann@tu-dresden.de)
*
Technische Universität Dresden // ‡Centre for Tactile Internet with Human-in-the-Loop (CeTI)
DOI 10.5281/zenodo.6963426 connect.relational-rags.eu
Context and Problem: Specifying Causal Connections for models@run.time

•Model-driven enginering reduces complexity during development of self-adaptive systems
•Models@run.time enables using models during runtime
•Connecting models to different external systems involves manual work
• easily, with a DSL

RQ1 How to send and receive updates of a
runtime model or any part of it to/from an add/refine computations Structure
revisit structure
external system? (Grammar)
RQ2 How to manage these updates
automatically and efficiently? revisit structure
❌ No connections to other systems possible
✅ Declarative specification ⬋
⬊ (only for simple tokens [Sch+20])
✅ Efficient execution using incremental evaluation Computation add/refine Semantics
(Attributes) semantics (Connections)
Solution: Self-Adaptive Systems with JastAdd [HM03] + Relational RAGs [Mey+20] + RagConnect
• efficiently, exploiting RAG technology

Connect Specification (Semantics) Grammar (Structure) AST as model@run.time
receive WorldModelB.MyScene using ParseScene, ConvertScene ; ① WorldModelB ::= Region* Robot* ExecutedOperation:Operation* send port Region OtherScene [0] LogicalScene
receive indexed WorldModelB.OtherScene ; ① [MyScene:Scene] OtherScene:LogicalScene*
receive port
name = "A"
WorldModelB
receive Robot.CurrentPosition ; // default mapping used /NextOperation:Operation/ ; ②

receive indexed with add WorldModelB.ExecutedOperation using
ParseCommand, ConvertCommand ;
Robot ::= <Name:String> <CurrentPosition> ;
send WorldModelB.NextOperation using PrintOperation ; ② PickAndPlace ;

rel PickAndPlace.RobotToExecute? -> Robot ; Robot
send Robot.myPosition(String) ;
rel PickAndPlace.ObjectToPick -> LogicalMovableObject ; OwnedCollaborationZoneNames
RobotToExecute
PrintOperation maps Operation op to byte[] {: rel PickAndPlace.TargetLocation -> DropOffLocation ; CurrentPosition

byte[] result = op.toProtobufByteArray(); // embedded java
if (result == null) { reject(); } // code for the // some of the nonterminals shared by both sites
Scene ::= DropOffLocation* MovableObject* RobotObject* ;
return result; // mapping content
LogicalScene ::= LogicalRegion* LogicalMovableObject* ;
:} CanReachObjectOfInterest MovableObject
Region ::= <Name:String> <LocationNames> ; LogicalScene
DropOffLocation
CollaborationZone : DropOffLocation ; ObjectName = "blue2" Name = "blue2"
RagConnect
Driver Code
⟨port⟩ ::= ⟨direction⟩ ⟨options⟩ ⟨target⟩ using ⟨mappings⟩ ; WorldModelB model = new WorldModelB();
model.ragconnectResetEvaluationCounter();
⟨direction⟩ ::= send | receive model.addOtherScene(new LogicalScene());
for (String topic : config.forB.topicsSceneUpdate) {

⟨options⟩ ::= ϵ | indexed | with add | indexed with add Attributes (Computation) model.connectMyScene(mqttUri(topic, config));
}
⟨target⟩ ::= ⟨nt-name⟩ syn Operation WorldModelB.getNextOperation() { ② model.connectOtherScene(mqttUri("place-a/logical/update", config), 0); ①
| ⟨nt-name⟩ . ⟨child-name⟩ if (diffToOperations().getNumChild() == 0) { model.connectNextOperation(mqttUri(config.forB.topicCommand, config), false);
model.connectExecutedOperation(mqttUri(config.forB.topicCommand, config));
②
return errorNoOperationComputed();
| ⟨nt-name⟩ . ⟨attr-name⟩ ( ⟨type-name⟩ )
}
⟨mappings⟩ ::= ⟨mapping-name⟩ , ⟨mappings⟩ | ⟨mapping-name⟩
for (Operation op : diffToOperations()) { JastAdd
Robot executingRobot = op.getRobotToExecute();
⟨mapping⟩ ::=
if (!op.isErrorOperation()) { uses
if (op.equals(lastOperationFor(executingRobot))) {
⟨mapping-name⟩ maps ⟨type-use⟩ ⟨ident⟩ to return errorDuplicateOperation();
⟨type-use⟩ {: ⟨mapping-content⟩ :} }
return op; Generated Code
⟨type-use⟩ ::= ⟨type-name⟩ | ⟨array-type-name⟩ [] }
} LOC of all Case Studies
return errorNoExecutableOperation();
es
ar
Support for Different Parts of the Grammar Using Send
ut
t
}
ec
ib
and Receive Ports. Legend: Full support, G
# Partly
nn
va
tr
ra
Co
At
Ja
supported, # Not supported, / not applicable Case Study Part
G
Element of a grammar Send Receive Connection Aspect Common (manual) 15 255 52 318
Message Receive Apply yes
Update Site A (manual) 1 17 2 136
Terminal message mappings
Different
value?
port
target Site A (generated) 31 2570 11 042
Main
We showed this
Non-Terminal (Context-free) if rejected Site B (manual) 29 668 136 350
no
Non-Terminal (Single) Site B (generated) 69 5443 23 075
• in a robot coordination case study

Non-Terminal (List) Main (manual) 15 261 8 144
Non-Terminal (Optional) Apply
Coordinator
Port target
has changed
Recompute
port target Different yes Send
message
mappings
Relation G
# # value?
Attribute / Recognized by incremental if rejected no

evaluation feature of JastAdd Message
Evaluation: Collaborative, Teaching-Based Robotic Cells (Case study 1) and Coordinator (Case study 2)
← Conceptual view of the two robots (site B) moving

Site A Site B objects to desired locations defined by an
Site A Site B
R1 operator (site A) MQTT
2 Broker
R1 R2
• . . . for solving a cooperative task

4
Two Franka Emica robot arms (site B) move

1 R3
3
Collabo-
ration
⬋ one object each to a new location in parallel.
R2
1 B1 Robot Zone B1 Robot
Operator Operator
Controller Controller
Scene Scene
Controller A R3 Controller A
2 3 4 B2 Robot B2 Robot
command
scene
Controller Controller
command
scene
scene
RAG command status RAG
Model A status Coordinator command Model B
Conceptual view on the coordinator starting →
• . . . and for coordinating the runtime components.

MovableObject Region Component Causal Connection Port Port Target components according to component dependencies Component Start Strategy Causal Connection Port Port Target
Our approach uses models based on relational RAGs and RagConnect enables adding efficient connections to all possible parts of a given model specification.
References
[HM03] Görel Hedin and Eva Magnusson. “JastAdd—an Aspect-Oriented Compiler Construction System”. In: Science of Computer Programming [Sch+20] René Schöne et al. “Connecting Conceptual Models Using Relational Reference Attribute Grammars”. In: Proceedings of the 23rd ACM/
47.1 (Apr. 1, 2003). IEEE International Conference on Model Driven Engineering Languages and Systems: Companion Proceedings. New York, NY, USA: ACM, Oct.
[Mey+20] Johannes Mey et al. “Relational Reference Attribute Grammars: Improving Continuous Model Validation”. In: Journal of Computer 16, 2020, pp. 1–11.
Languages 57 (2020).
Member in the network of:

This work is partly supported by the German Research Foundation (DFG) as
part of Germany's Excellence Strategy -- EXC 2050/1 -- Project 390696704
-- "Centre for Tactile Internet with Human-in-the-Loop" (CeTI), by the
Federal Ministry of Education and Research of Germany in the programme
of "Souverän. Digital. Vernetzt." Joint project 6G-life, project identification
number: 16KISK001K, and using tax money based on the budget approved
by the Saxon state parliament for the project "PROSPER".
We provide the missing concepts to use RAGs for Models@run.time!
References I
[Hed00] Görel Hedin. “Reference attributed grammars”. In: Informatica (Slovenia) 24.3
(2000), pp. 301–317. URL: http://citeseerx.ist.psu.edu/viewdoc/download?doi=
10.1.1.108.8792&rep=rep1&type=pdf (visited on 08/03/2015).
[HM03] Görel Hedin and Eva Magnusson. “JastAdd: an aspect-oriented compiler
construction system”. In: Science of Computer Programming. Special Issue on
Language Descriptions, Tools and Applications (L DTA’01) 47.1 (Apr. 2003),
pp. 37–58. ISSN: 0167-6423. DOI: 10.1016/S0167-6423(02)00109-0. URL:
http://www.sciencedirect.com/science/article/pii/S0167642302001090 (visited on
10/01/2019).
[Knu68] Donald E. Knuth. “Semantics of context-free languages”. In: Mathematical systems
theory 2.2 (1968), pp. 127–145. URL:
http://link.springer.com/article/10.1007/BF01692511 (visited on 08/03/2015).
References II
[Mey+20] Johannes Mey et al. “Relational Reference Attribute Grammars: Improving

Continuous Model Validation”. en. In: Journal of Computer Languages 57 (Jan. 2020),
p. 21. ISSN: 2590-1184. DOI: 10.1016/j.cola.2019.100940. URL:
http://www.sciencedirect.com/science/article/pii/S2590118419300656 (visited on
02/14/2020).
[Wey19] Danny Weyns. “Software Engineering of Self-adaptive Systems”. en. In: Handbook
of Software Engineering. Cham: Springer International Publishing, 2019,
pp. 399–443. ISBN: 978-3-030-00262-6. URL:
https://doi.org/10.1007/978-3-030-00262-6_11.
RagConnect DSL // Grammar
⟨port⟩ ::= ⟨direction⟩ ⟨options⟩ ⟨target⟩ using ⟨mappings⟩ ;
⟨direction⟩ ::= send | receive
⟨options⟩ ::= ϵ | indexed | with add | indexed with add
⟨target⟩ ::= ⟨nt-name⟩

| ⟨nt-name⟩ . ⟨child-name⟩
| ⟨nt-name⟩ . ⟨attr-name⟩ ( ⟨type-name⟩ )
⟨mappings⟩ ::= ⟨mapping-name⟩ , ⟨mappings⟩ | ⟨mapping-name⟩

⟨mapping⟩ ::=
⟨mapping-name⟩ maps ⟨type-use⟩ ⟨ident⟩ to
⟨type-use⟩ {: ⟨mapping-content⟩ :}
⟨type-use⟩ ::= ⟨type-name⟩ | ⟨array-type-name⟩ []

Incremental Causal Connection For Self-Adaptive Systems Based On Relational Reference Attribute Grammars

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Incremental Causal Connection For Self-Adaptive Systems Based On Relational Reference Attribute Grammars

Uploaded by

Copyright:

Available Formats

René Schöne, Johannes Mey, Sebastian Ebert, Sebastian Götz, Uwe Aßmann

Incremental Causal Connection for Self-Adaptive

MovableObject Region Component Causal Connection Port Port Target

MovableObject Region Component Causal Connection Port Port Target

What if: we use RAGs as technological space for the model?

Connections from Managing System

Self-adaptive System Speciﬁcations of Managing System

Connections from Managing System

Connections from Managing System

Attribute Grammars specify semantics/behaviour for a tree of a context-free grammar.

We use the JastAdd object-oriented Reference AG system [HM03].

– Containment relations form overlay tree

WorldModel ::= [MyScene:Scene]

abstract Object ::= <Name:String> Position;

rel PickAndPlace.ObjectToPick -> MovableObject;

WorldModel ::= [MyScene:Scene]

abstract Object ::= <Name:String> Position;

rel PickAndPlace.ObjectToPick -> MovableObject;

WorldModel ::= [MyScene:Scene]

abstract Object ::= <Name:String> Position;

rel PickAndPlace.ObjectToPick -> MovableObject;

WorldModel ::= [MyScene:Scene]

abstract Object ::= <Name:String> Position;

rel PickAndPlace.ObjectToPick -> MovableObject;

WorldModel ::= [MyScene:Scene]

abstract Object ::= <Name:String> Position;

rel PickAndPlace.ObjectToPick -> MovableObject;

WorldModel ::= [MyScene:Scene]

abstract Object ::= <Name:String> Position;

rel PickAndPlace.ObjectToPick -> MovableObject;

WorldModel ::= [MyScene:Scene]

abstract Object ::= <Name:String> Position;

rel PickAndPlace.ObjectToPick -> MovableObject;

WorldModel ::= [MyScene:Scene]

abstract Object ::= <Name:String> Position;

rel PickAndPlace.ObjectToPick -> MovableObject;

add/reﬁne computations Structure

add/reﬁne computations Structure

Computation add/reﬁne Semantics

Reachability [0] ObjectToPick

CanReachObjectOfInterest MovableObject LogicalScene

Reachability [0] ObjectToPick

CanReachObjectOfInterest MovableObject LogicalScene

Recognized by incremental if rejected no

LOC of all Case Studies

LOC of all Case Studies

This work presents

We provide the missing concepts to use RAGs for Models@run.time!

This work presents

Context and Problem: Specifying Causal Connections for models@run.time

• easily, with a DSL

• eﬃciently, exploiting RAG technology

receive Robot.CurrentPosition ; // default mapping used /NextOperation:Operation/ ; ②

send WorldModelB.NextOperation using PrintOperation ; ② PickAndPlace ;

PrintOperation maps Operation op to byte[] {: rel PickAndPlace.TargetLocation -> DropOffLocation ; CurrentPosition

for (String topic : config.forB.topicsSceneUpdate) {

• in a robot coordination case study

Attribute / Recognized by incremental if rejected no

1st phase: Speciﬁcation 2nd phase: Generation 3rd phase: Usage

← Conceptual view of the two robots (site B) moving

• . . . for solving a cooperative task

Two Franka Emica robot arms (site B) move

• . . . and for coordinating the runtime components.