Exercise 5: Cognitive Agents in Hypermedia Environments

A template for an application implemented with the JaCaMo 0.9 framework for programming Multi-Agent Systems (MAS).

Table of Contents

• Task 1
• Task 2
• Task 3
• Register as a user to leubot1
• Programming JaCaMo applications in Atom
• How to run the project

Task 1

Task 1 requires implementing (part of) a tutorial for programming four agents that will print different hello world messages.

Follow the instructions of the Hello JaCaMo - Part II (agent) tutorial. Update the files sample_agent.asl and task1.jcm based on the instructions.

HINTS:

• Run the Gradle task task1 based on the instructions to observe the behavior of your implementation.

Task 2

Task2 requires updating the program of a detector agent for enabling it to infer knowledge based on the current state of the world.

A detector agent uses a simulated service detector1 for detecting blocks whithin a workspace named "lab". Based on the notifications that the agent receives from the service, it constructs beliefs of the form positioned(BlockName, X, Y, Z)[certainty(Cert)], e.g. positioned("A", 100, 150, 0)[certainty(0.9)]. Such a belief denotes that the agent believes that block "A" is positioned in (100, 150, 0) with a degree of certainty 0.9. Based on its beliefs, the agent follows Prolog-like rules to infer whether a block is positioned on top of another block, and whether a block is positioned on the table within the workspace.

STEPS:

• Step 1: Update the rule on(Block1, Block2) to enable the agent to infer whether a Block1 is detected to be positioned on top of a Block2.
• Step 2: Update the rule onTable(Block) to enable the agent to infer whether a Block is detected to be positioned on the table.

HINTS:

• Observe how the similar rule clear(Block) is implemented for inferring whether nothing is positioned on top of a Block.
• For implementing the rules, take into consideration the degree of certainty of the agent's beliefs. The agent should not infer anything if any related belief has a degree of certainty < 0.5.
• Run the Gradle task task2 based on the instructions to test your implementation.

Task 3

Task3 requires updating the program of a mover agent for planning the solution of the blocks-world problem in the lab workspace.

A mover agent uses the following:

• A simulated service detector1 for detecting blocks whithin the lab workspace.
• A robotic arm leubot1 for moving blocks within the lab workspace.

By using detector1 and its rules (implemented in Task 1), the agent believes (as described in Task 1), at some point in time, that the state of the world is as follows:

Based on its beliefs about the state, the agent attempts to organize the blocks "A", "B", "C" using the robotic arm leubot1, so that the state of the world looks as follows:

After organizing the blocks, the agent prints its current beliefs that should have been updated based on the notifications of detector1 and should correspond to the updated state of the world.

STEPS:

• Step 1: Register as a user to leubot1, so that the agent can use the robotic arm. Follow the instructions for acquiring an API key, and then update the agent's belief apikey accordingly. Alternatively, if you are running the application on dry run (i.e. without executing the requests to leubot1), simply uncomment the existing belief. This belief is required for the agent to start (see plan start).
• Step 2: Update the rules on(Block1, Block2) and onTable(Block) based on the work that you implemented in Task 2. These rules are useful for the agent to infer beliefs required for organizing the blocks.
• Step 3: Update the context and body of the agent's plan organize(Block1, Block2, Block3) to enable the agent to organize Block1, Block2, and Block3 based on the blocks-world problem. Use the agent's beliefs (e.g. beliefs inferred through the agent's rules) to implement the context of the plan. Use the plans unstack(Block1, Block2), pickUp(Block), putDown(Block), and stack(Block1, Block2) to implement the plan based on sub-goals.
• Step 4: Update the context and body of the agent's plan putDown(Block) to enable the agent to put a Block on the table by using leubot1.
• Step 5: Update the context and body of the agent's plan pickUp(Block) to enable the agent to pick a Block from the table by using leubot1.
• Step 6: Update the context and body of the agent's plan stack(Block1, Block2) to enable the agent to stack a Block1 on Block2 by using leubot1.

HINTS:

• For implementing Tasks 4-6, observe how the similar plan unstack(Block1, Block2) is implemented for enabling the agent to ustack a Block1 from a Block2 by using leubot1.
• For implementing Tasks 3-6, avoid updating by yourselves any agent's beliefs that should be inferred through the agent's rules. Such beliefs should be automatically updated through the rules whenever detector1 notifies the agent about a change in the state of the world. For example, avoid making a belief addition +on(Block1,Block2) on the plan body of stack(Block1,Block2). Instead, concentrate on implementing a body that triggers the execution of existing plans (e.g. !unstack(Block1, Block2) etc. in Step 3, and !move(X,Y,Z), !release etc. in Steps 3-6), and that updates the agent's beliefs about leubot1 (e.g. gripperEmpty etc.).
• Run the Gradle task task3 or task3dryrun based on the instructions to observe the behavior of your implementation.

Register as a user to leubot1

For registering as a user and exploit the API of leubot1 v1.3.4, you need to send an HTTP POST request to https://api.interactions.ics.unisg.ch/leubot1/v1.3.4/user with your name and email in a JSON payload. The last element of the response header Location is the API key that is required for interacting with leubot1.

For example, if you are using cURL, you can run a similar command:

curl -v https://api.interactions.ics.unisg.ch/leubot1/v1.3.4/user -H 'Content-Type: application/json' --data '{"name": "Example Name", "email": "[email protected]"}'

Upon execution, you should receive a similar response:

HTTP/1.1 201 Created
< Server: nginx/1.14.1
< Date: Tue, 22 Mar 2022 21:07:32 GMT
< Content-Type: application/json; charset=UTF-8
< Content-Length: 0
< Connection: keep-alive
< Access-Control-Allow-Credentials: true
< Access-Control-Allow-Headers: *
< Access-Control-Allow-Methods: GET,HEAD,OPTIONS,POST
< Access-Control-Allow-Origin: *
< Location: https://api.interactions.ics.unisg.ch/leubot1/v1.3.4/user/047820620ee091c7301b3fceed54d918

Here, 047820620ee091c7301b3fceed54d918 is the valid API key.

The W3C Web of Things Thing Description of leubot1 is available here.

Programming JaCaMo applications in Atom

The Atom editor supports syntax highlighting for programming JaCaMo applications. Suggested packages:

• language-agentspeak: syntax highlighting for agent files (.asl) with the AgentSpeak language [1].
• language-jcm: syntax highlighting for the configurations files (.jcm) of JaCaMo MAS applications.
• ide-java and atom-ide-ui: syntax highlighting for Java files (.java) used for the provided implementations of detector1 and leubot1.

To install any of the suggested packages, open your Atom editor, and Packages -> Settings View -> Install Packages/Themes, and search and Install your preferred package.

[1] Bordini, R. H., Hübner, J. F., & Wooldridge, M. (2007). Programming multi-agent systems in AgentSpeak using Jason. John Wiley & Sons.

How to run the project

Run with Gradle 7.4:

• MacOS and Linux: run the following commands
• Windows: replace ./gradlew with gradlew.bat

For Task 1:

./gradlew task1

For Task 2:

./gradlew task2

For Task 3 by printing and executing the requests to leubot1):

./gradlew task3

For Task 3 by only printing the requests to leubot1):

./gradlew task3dryrun