CS823 Assignment 1: Part 1: Classical Planning and Part 2: Temporal Planning

CS823 Assignment 1: Classical and Temporal Planning

Overview

This assignment is worth 25% of the mark for this module. You will apply both classical and temporal planning techniques in PDDL to model and solve package delivery problems. The first half of the assignment will involve classical planning without time-bound constraints, while the second half will focus on temporal planning with durative actions and temporal constraints.

Objectives

1. To gain hands-on experience in classical and temporal planning in package delivery.
2. To use predicates, actions, and constraints to model package delivery in classical and temporal planning contexts.
3. To understand the differences between classical and temporal planning and the unique challenges they pose.

Part 1: Classical Planning

In this part, you will use classical planning to design domain and problem for package delivery without temporal constraints. you should focus on creating a functional and efficient model for delivering packages using vans across locations.

1.Design a Classical Domain File ( domain_classical.pddl )

• Define a domain file with classical (non-durative) actions for  

• walk the driver from a location to another
• board the driver into a van at some location
• unboard the driver from a van at some location
• the driver drive a van from a location into another
• the driver load a package into a van at some location
• the driver unload a package from a van at some location

• Specify the actions to operate without temporal constraints but with logical dependencies, such as van capacities, and loading/unloading conditions.
• Ensure that the domain includes relevant predicates to check conditions for each action (e.g., package is at location, van is empty, etc.).

2.Design a Classical Problem File ( problem_classical.pddl )

• Create a problem file that applies the classical domain to a scenario involving multiple packages, vans, and locations.
• Define initial states, goals, and logical constraints (e.g., locations where vans or packages begin, goal conditions for completed deliveries).
• Test the problem using a classical planner and document the results, focusing on efficiency and completeness.

Part 2: Temporal Planning

The second part of the assignment will focus on temporal planning, where you add temporal constraints to the problem, such as durative actions and time windows. This part should build on the classical model but include advanced, time-bound planning.

1.Design a Temporal Domain File ( domain_temporal.pddl )

• Extend the classical domain to incorporate durative actions.
• Define realistic temporal constraints, such as durations for each action, capacity constraints, and dependencies that require actions to occur in specific sequences.
• Ensure that actions and predicates accurately reflect the need for timing and concurrency
  (e.g., the ability to transport packages while other actions occur in parallel).

2.Design a Temporal Problem File ( problem_temporal.pddl )

• Create a temporal problem file that introduces new challenges, such as specific time windows for deliveries or restricted access times for locations.
• Define initial states, temporal constraints, and goal conditions that match a realistic

scenario.

Test the temporal problem in a temporal planner, document the results, and provide insights into the impact of durative actions and constraints.

Unit Test - Classical Planning

Initial state

• Route Connections  loc1 <-> loc2
• driver1 at loc1
• van1 empty and at loc2  pack1 is at loc2

Goal state

• driver1 at loc1  van1 at loc2
• pack1 at loc1

0.00000: (walk driver1 loc1 loc2) [0.00100]
0.00100: (board_van van1 driver1 loc2) [0.00100]
0.00200: (load_package van1 driver1 pack1 loc2) [0.00100] 0.00300: (drive_van van1 driver1 loc2 loc1) [0.00100] 0.00400: (unload_package 
van1 driver1 pack1 loc1) [0.00100] 0.00500: (drive_van van1 driver1 loc1 loc2) [0.00100] 0.00600: (unboard_van van1 driver1 loc2) [0.00100]
0.00700: (walk driver1 loc2 loc1) [0.00100] Plan found!

Submission

1.Classical Domain and Problem Files ( domain_classical.pddl and problem_classical.pddl ).
2.Temporal Domain and Problem Files ( domain_temporal.pddl and problem_temporal.pddl ).

Marking Scheme

Use of Generative AI (GenAI) Tools – Disclaimer and Guidance

Students may make limited use of Generative AI tools (e.g., ChatGPT, Gemini, Copilot) to support their work on this assignment only for general learning support and code debugging assistance. However, any use of such tools must be appropriately acknowledged in your submission.

You must not use GenAI to generate entire code blocks, explanations, or comments without critical modification and full understanding. Submissions found to include uncredited or overly reliant GenAI- generated content may be subject to academic misconduct procedures.

If you have used any GenAI tool, you must include a brief "GenAI Acknowledgement" in your submitted PDDL files as code comments, stating:
The tool(s) used (e.g., ChatGPT, Copilot)

What you used it for (e.g., debugging syntax errors, explaining concepts)  A short reflection on how it informed your understanding
This aligns with the University’s policy on ethical and transparent use of emerging technologies in assessments.