Project 9: STRIPS Planner (AI Planning)

Build a STRIPS-style planner that finds action sequences to achieve goals.

Quick Reference

Attribute Value
Difficulty Level 4: Expert
Time Estimate 12-20 hours
Language Python
Prerequisites Search algorithms, logic basics
Key Topics planning, operators, goal states

1. Learning Objectives

By completing this project, you will:

  1. Represent states and actions in STRIPS form.
  2. Implement forward planning search.
  3. Handle preconditions and effects.
  4. Generate valid action plans.
  5. Measure plan length and search cost.

2. Theoretical Foundation

2.1 STRIPS Planning

STRIPS planners define actions with preconditions and effects to move between states.

3. Project Specification

3.1 What You Will Build

A planner that takes an initial state and goal state and outputs a sequence of actions.

3.2 Functional Requirements

  1. State representation with predicates.
  2. Action operators with preconditions/effects.
  3. Search algorithm (BFS/A*).
  4. Plan output with action sequence.
  5. Metrics for plan length and cost.

3.3 Non-Functional Requirements

  • Deterministic outputs for same inputs.
  • Configurable action sets.
  • Readable plan output.

4. Solution Architecture

4.1 Components

Component Responsibility
State Model Represent predicates
Operators Define actions
Planner Search for plan
Reporter Plan + metrics

5. Implementation Guide

5.1 Project Structure

SYMBOLIC_AI_AND_EXPERT_SYSTEMS_MASTERY/P09-strips/
├── src/
│   ├── state.py
│   ├── operators.py
│   ├── planner.py
│   └── report.py

5.2 Implementation Phases

Phase 1: State + operators (4-6h)

  • Define predicates and actions.
  • Checkpoint: actions apply correctly.

Phase 2: Planner (4-8h)

  • Implement forward search.
  • Checkpoint: plan found for simple goals.

Phase 3: Metrics (3-6h)

  • Track plan length and cost.
  • Checkpoint: report generated.

6. Testing Strategy

6.1 Test Categories

Category Purpose Examples
Unit operators preconditions/effects
Integration planning valid plan output
Regression metrics stable plan lengths

6.2 Critical Test Cases

  1. Planner finds plan for known problems.
  2. Invalid actions are not applied.
  3. Plan achieves goal state.

7. Common Pitfalls & Debugging

Pitfall Symptom Fix
State explosion slow search add heuristics
Invalid actions broken plans validate preconditions
Non-termination infinite search add depth limits

8. Extensions & Challenges

Beginner

  • Add simple planning domains.
  • Add plan visualization.

Intermediate

  • Add heuristic search (A*).
  • Add plan caching.

Advanced

  • Add partial-order planning.
  • Compare with PDDL planners.

9. Real-World Connections

  • Robotics uses planning systems.
  • Workflow automation relies on action planning.

10. Resources

  • STRIPS planning references
  • AI planning textbooks

11. Self-Assessment Checklist

  • I can represent actions and states.
  • I can generate valid plans.
  • I can measure planning cost.

12. Submission / Completion Criteria

Minimum Completion:

  • STRIPS planner with action sequences

Full Completion:

  • Metrics + domain examples

Excellence:

  • Heuristic planning or PDDL integration

This guide was generated from project_based_ideas/AI_AGENTS_LLM_RAG/SYMBOLIC_AI_AND_EXPERT_SYSTEMS_MASTERY.md.