"""
Pattern 163: Spatial Reasoning
Description:
The Spatial Reasoning pattern enables agents to understand and reason about spatial
relationships, positions, orientations, and geometric properties. This pattern helps
agents solve problems involving physical space, layouts, navigation, and spatial
configurations.
Components:
1. Coordinate System: Tracks positions and locations
2. Relationship Analyzer: Identifies spatial relationships (above, below, left, right, etc.)
3. Distance Calculator: Computes distances and proximity
4. Orientation Tracker: Manages directions and rotations
5. Spatial Query Engine: Answers spatial questions
6. Visualization Helper: Generates spatial descriptions
Use Cases:
- Navigation and pathfinding
- Layout design and optimization
- Object placement and arrangement
- Map understanding and interpretation
- Geometry problem solving
- Spatial puzzle solving
- Floor plan analysis
Benefits:
- Enables spatial understanding
- Supports geometric reasoning
- Facilitates navigation tasks
- Improves layout optimization
- Better scene understanding
Trade-offs:
- Complex coordinate management
- Requires clear spatial representation
- May need visualization capabilities
- Challenging for abstract spaces
- Limited by 2D/3D representation
LangChain Implementation:
Uses LLM for spatial relationship interpretation with structured coordinate
systems and distance calculations. Combines symbolic representation with
natural language understanding for spatial queries.
"""
import os
import math
from typing import List, Dict, Any, Tuple, Optional
from dataclasses import dataclass
from enum import Enum
from dotenv import load_dotenv
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
load_dotenv()
class Direction(Enum):
"""Cardinal and ordinal directions"""
NORTH = "north"
SOUTH = "south"
EAST = "east"
WEST = "west"
NORTHEAST = "northeast"
NORTHWEST = "northwest"
SOUTHEAST = "southeast"
SOUTHWEST = "southwest"
UP = "up"
DOWN = "down"
class SpatialRelation(Enum):
"""Types of spatial relationships"""
ABOVE = "above"
BELOW = "below"
LEFT_OF = "left of"
RIGHT_OF = "right of"
IN_FRONT_OF = "in front of"
BEHIND = "behind"
INSIDE = "inside"
OUTSIDE = "outside"
ADJACENT = "adjacent"
NEAR = "near"
FAR = "far"
BETWEEN = "between"
@dataclass
class SpatialObject:
"""Represents an object in space"""
name: str
x: float
y: float
z: float = 0.0
width: float = 1.0
height: float = 1.0
depth: float = 1.0
orientation: float = 0.0 # degrees
@dataclass
class SpatialQuery:
"""A spatial reasoning query"""
question: str
objects: List[SpatialObject]
context: str = ""
@dataclass
class SpatialAnalysis:
"""Result of spatial reasoning"""
answer: str
relationships: List[str]
distances: Dict[str, float]
reasoning: str
visualization: str
class SpatialReasoningAgent:
"""Agent that performs spatial reasoning tasks"""
def __init__(self, model_name: str = "gpt-3.5-turbo"):
"""Initialize the spatial reasoning agent"""
self.llm = ChatOpenAI(model=model_name, temperature=0)
# Prompts for different spatial reasoning tasks
self.relation_prompt = ChatPromptTemplate.from_messages([
("system", """You are a spatial reasoning expert. Analyze the positions
of objects and describe their spatial relationships clearly and accurately.
Consider distances, directions, and relative positions."""),
("user", """Objects and their positions:
{object_descriptions}
Question: {question}
Provide a clear answer based on the spatial relationships.""")
])
self.path_prompt = ChatPromptTemplate.from_messages([
("system", """You are a navigation expert. Given object positions,
determine optimal paths and directions."""),
("user", """Objects and their positions:
{object_descriptions}
Navigation task: {task}
Describe the path or navigation instructions.""")
])
self.relation_chain = self.relation_prompt | self.llm | StrOutputParser()
self.path_chain = self.path_prompt | self.llm | StrOutputParser()
def analyze_spatial_query(self, query: SpatialQuery) -> SpatialAnalysis:
"""Analyze a spatial query and return comprehensive analysis"""
# Calculate all pairwise distances
distances = self._calculate_distances(query.objects)
# Identify spatial relationships
relationships = self._identify_relationships(query.objects)
# Create object descriptions
object_descriptions = self._create_object_descriptions(query.objects)
# Get LLM reasoning
reasoning = self.relation_chain.invoke({
"object_descriptions": object_descriptions,
"question": query.question
})
# Create visualization
visualization = self._create_visualization(query.objects)
return SpatialAnalysis(
answer=reasoning,
relationships=relationships,
distances=distances,
reasoning=reasoning,
visualization=visualization
)
def find_path(self, start: SpatialObject, end: SpatialObject,
obstacles: List[SpatialObject]) -> Dict[str, Any]:
"""Find path from start to end avoiding obstacles"""
# Calculate direct distance
direct_distance = self._distance(start, end)
# Calculate direction
direction = self._get_direction(start, end)
# Check for obstacles in path
obstacles_in_path = self._check_obstacles(start, end, obstacles)
# Get navigation instructions
object_descriptions = self._create_object_descriptions(
[start, end] + obstacles
)
instructions = self.path_chain.invoke({
"object_descriptions": object_descriptions,
"task": f"Navigate from {start.name} to {end.name}"
})
return {
"direct_distance": direct_distance,
"direction": direction,
"obstacles_in_path": obstacles_in_path,
"instructions": instructions,
"path_clear": len(obstacles_in_path) == 0
}
def find_nearest(self, reference: SpatialObject,
candidates: List[SpatialObject]) -> Tuple[SpatialObject, float]:
"""Find the nearest object to reference from candidates"""
nearest = None
min_distance = float('inf')
for candidate in candidates:
distance = self._distance(reference, candidate)
if distance bool:
"""Check if a specific spatial relationship holds between two objects"""
if relation == SpatialRelation.ABOVE:
return obj1.z > obj2.z
elif relation == SpatialRelation.BELOW:
return obj1.z < obj2.z
elif relation == SpatialRelation.LEFT_OF:
return obj1.x obj2.x
elif relation == SpatialRelation.IN_FRONT_OF:
return obj1.y > obj2.y
elif relation == SpatialRelation.BEHIND:
return obj1.y < obj2.y
elif relation == SpatialRelation.NEAR:
distance = self._distance(obj1, obj2)
return distance 10.0 # Threshold for "far"
elif relation == SpatialRelation.ADJACENT:
distance = self._distance(obj1, obj2)
return distance List[SpatialObject]:
"""Get all objects within a spherical region"""
result = []
cx, cy, cz = center
for obj in objects:
distance = math.sqrt(
(obj.x - cx)**2 +
(obj.y - cy)**2 +
(obj.z - cz)**2
)
if distance float:
"""Calculate Euclidean distance between two objects"""
return math.sqrt(
(obj1.x - obj2.x)**2 +
(obj1.y - obj2.y)**2 +
(obj1.z - obj2.z)**2
)
def _calculate_distances(self, objects: List[SpatialObject]) -> Dict[str, float]:
"""Calculate all pairwise distances"""
distances = {}
for i, obj1 in enumerate(objects):
for obj2 in objects[i+1:]:
key = f"{obj1.name} to {obj2.name}"
distances[key] = self._distance(obj1, obj2)
return distances
def _identify_relationships(self, objects: List[SpatialObject]) -> List[str]:
"""Identify spatial relationships between objects"""
relationships = []
for i, obj1 in enumerate(objects):
for obj2 in objects[i+1:]:
# Check various relationships
if self.check_relationship(obj1, obj2, SpatialRelation.ABOVE):
relationships.append(f"{obj1.name} is above {obj2.name}")
if self.check_relationship(obj1, obj2, SpatialRelation.BELOW):
relationships.append(f"{obj1.name} is below {obj2.name}")
if self.check_relationship(obj1, obj2, SpatialRelation.LEFT_OF):
relationships.append(f"{obj1.name} is left of {obj2.name}")
if self.check_relationship(obj1, obj2, SpatialRelation.RIGHT_OF):
relationships.append(f"{obj1.name} is right of {obj2.name}")
if self.check_relationship(obj1, obj2, SpatialRelation.NEAR):
relationships.append(f"{obj1.name} is near {obj2.name}")
if self.check_relationship(obj1, obj2, SpatialRelation.ADJACENT):
relationships.append(f"{obj1.name} is adjacent to {obj2.name}")
return relationships
def _get_direction(self, from_obj: SpatialObject,
to_obj: SpatialObject) -> Direction:
"""Determine cardinal direction from one object to another"""
dx = to_obj.x - from_obj.x
dy = to_obj.y - from_obj.y
angle = math.atan2(dy, dx) * 180 / math.pi
# Normalize to 0-360
if angle < 0:
angle += 360
# Determine direction
if 22.5 <= angle < 67.5:
return Direction.NORTHEAST
elif 67.5 <= angle < 112.5:
return Direction.NORTH
elif 112.5 <= angle < 157.5:
return Direction.NORTHWEST
elif 157.5 <= angle < 202.5:
return Direction.WEST
elif 202.5 <= angle < 247.5:
return Direction.SOUTHWEST
elif 247.5 <= angle < 292.5:
return Direction.SOUTH
elif 292.5 <= angle List[str]:
"""Check which obstacles are in the path between start and end"""
obstacles_in_path = []
for obstacle in obstacles:
# Simple line intersection check
if self._is_near_line(start, end, obstacle):
obstacles_in_path.append(obstacle.name)
return obstacles_in_path
def _is_near_line(self, start: SpatialObject, end: SpatialObject,
point: SpatialObject, threshold: float = 1.0) -> bool:
"""Check if a point is near the line between start and end"""
# Calculate perpendicular distance from point to line
x1, y1 = start.x, start.y
x2, y2 = end.x, end.y
x0, y0 = point.x, point.y
numerator = abs((y2-y1)*x0 - (x2-x1)*y0 + x2*y1 - y2*x1)
denominator = math.sqrt((y2-y1)**2 + (x2-x1)**2)
if denominator == 0:
return False
distance = numerator / denominator
# Also check if point is between start and end
between = (min(x1, x2) <= x0 <= max(x1, x2) and
min(y1, y2) <= y0 <= max(y1, y2))
return distance str:
"""Create textual descriptions of object positions"""
descriptions = []
for obj in objects:
desc = f"{obj.name}: position ({obj.x:.1f}, {obj.y:.1f}, {obj.z:.1f})"
if obj.width != 1.0 or obj.height != 1.0:
desc += f", size ({obj.width:.1f}x{obj.height:.1f}x{obj.depth:.1f})"
descriptions.append(desc)
return "\n".join(descriptions)
def _create_visualization(self, objects: List[SpatialObject]) -> str:
"""Create ASCII visualization of object positions"""
if not objects:
return "No objects to visualize"
# Find bounds
min_x = min(obj.x for obj in objects) - 2
max_x = max(obj.x for obj in objects) + 2
min_y = min(obj.y for obj in objects) - 2
max_y = max(obj.y for obj in objects) + 2
# Create grid
width = int(max_x - min_x) + 1
height = int(max_y - min_y) + 1
grid = [[' ' for _ in range(width)] for _ in range(height)]
# Place objects
for i, obj in enumerate(objects):
x = int(obj.x - min_x)
y = int(obj.y - min_y)
if 0 <= x < width and 0 <= y < height:
grid[height - 1 - y][x] = chr(65 + i) # A, B, C, ...
# Create visualization string
vis = "\n" + "\n".join("".join(row) for row in grid)
vis += "\n\nLegend: " + ", ".join(
f"{chr(65+i)}={obj.name}" for i, obj in enumerate(objects)
)
return vis
def demonstrate_spatial_reasoning():
"""Demonstrate spatial reasoning capabilities"""
print("=" * 80)
print("SPATIAL REASONING PATTERN DEMONSTRATION")
print("=" * 80)
agent = SpatialReasoningAgent()
# Example 1: Room layout analysis
print("\n" + "=" * 80)
print("Example 1: Room Layout Analysis")
print("=" * 80)
room_objects = [
SpatialObject("Table", 5, 5, 0, 2, 1, 3),
SpatialObject("Chair1", 4, 5, 0, 1, 1, 1),
SpatialObject("Chair2", 6, 5, 0, 1, 1, 1),
SpatialObject("Lamp", 5, 5, 2, 0.5, 0.5, 1),
SpatialObject("Door", 0, 5, 0, 1, 2, 0.1),
SpatialObject("Window", 10, 5, 1, 2, 1, 0.1)
]
query = SpatialQuery(
question="What is the spatial arrangement of furniture in the room?",
objects=room_objects,
context="Living room layout"
)
analysis = agent.analyze_spatial_query(query)
print(f"\nQuery: {query.question}")
print(f"\nAnswer: {analysis.answer}")
print(f"\nKey Relationships:")
for rel in analysis.relationships[:5]:
print(f" - {rel}")
print(f"\nVisualization:{analysis.visualization}")
# Example 2: Navigation pathfinding
print("\n" + "=" * 80)
print("Example 2: Navigation Pathfinding")
print("=" * 80)
start = SpatialObject("Start", 0, 0, 0)
end = SpatialObject("Goal", 10, 10, 0)
obstacles = [
SpatialObject("Obstacle1", 5, 5, 0, 2, 2, 1),
SpatialObject("Obstacle2", 7, 3, 0, 1, 1, 1)
]
path_result = agent.find_path(start, end, obstacles)
print(f"\nNavigation from {start.name} to {end.name}")
print(f"Direct distance: {path_result['direct_distance']:.2f} units")
print(f"Direction: {path_result['direction'].value}")
print(f"Path clear: {path_result['path_clear']}")
if path_result['obstacles_in_path']:
print(f"Obstacles in path: {', '.join(path_result['obstacles_in_path'])}")
print(f"\nInstructions:\n{path_result['instructions']}")
# Example 3: Nearest neighbor search
print("\n" + "=" * 80)
print("Example 3: Nearest Neighbor Search")
print("=" * 80)
reference = SpatialObject("Home", 5, 5, 0)
stores = [
SpatialObject("Grocery Store", 3, 4, 0),
SpatialObject("Pharmacy", 8, 6, 0),
SpatialObject("Coffee Shop", 5, 7, 0),
SpatialObject("Bookstore", 10, 10, 0)
]
nearest, distance = agent.find_nearest(reference, stores)
print(f"\nFinding nearest store to {reference.name}")
print(f"Nearest: {nearest.name}")
print(f"Distance: {distance:.2f} units")
print("\nAll distances:")
for store in stores:
dist = agent._distance(reference, store)
direction = agent._get_direction(reference, store)
print(f" {store.name}: {dist:.2f} units {direction.value}")
# Example 4: Spatial relationship checking
print("\n" + "=" * 80)
print("Example 4: Spatial Relationship Checking")
print("=" * 80)
ceiling_light = SpatialObject("Ceiling Light", 5, 5, 8)
floor_lamp = SpatialObject("Floor Lamp", 5, 5, 0)
couch = SpatialObject("Couch", 3, 5, 0)
relations_to_check = [
(ceiling_light, floor_lamp, SpatialRelation.ABOVE),
(floor_lamp, ceiling_light, SpatialRelation.BELOW),
(couch, floor_lamp, SpatialRelation.NEAR),
(couch, floor_lamp, SpatialRelation.ADJACENT)
]
print("\nChecking spatial relationships:")
for obj1, obj2, relation in relations_to_check:
holds = agent.check_relationship(obj1, obj2, relation)
print(f" {obj1.name} is {relation.value} {obj2.name}: {holds}")
# Example 5: Regional search
print("\n" + "=" * 80)
print("Example 5: Regional Search")
print("=" * 80)
all_objects = [
SpatialObject("Object1", 2, 2, 0),
SpatialObject("Object2", 5, 5, 0),
SpatialObject("Object3", 8, 8, 0),
SpatialObject("Object4", 3, 7, 0),
SpatialObject("Object5", 10, 2, 0)
]
search_center = (5, 5, 0)
search_radius = 4
objects_in_region = agent.get_objects_in_region(
all_objects, search_center, search_radius
)
print(f"\nSearching for objects within {search_radius} units of {search_center}")
print(f"Found {len(objects_in_region)} objects:")
for obj in objects_in_region:
dist = math.sqrt(
(obj.x - search_center[0])**2 +
(obj.y - search_center[1])**2 +
(obj.z - search_center[2])**2
)
print(f" {obj.name} at ({obj.x}, {obj.y}, {obj.z}) - {dist:.2f} units away")
# Example 6: Complex spatial query
print("\n" + "=" * 80)
print("Example 6: Complex Spatial Query")
print("=" * 80)
city_objects = [
SpatialObject("City Hall", 5, 5, 0, 3, 3, 10),
SpatialObject("Park", 8, 8, 0, 5, 5, 0),
SpatialObject("School", 2, 8, 0, 2, 2, 5),
SpatialObject("Hospital", 9, 2, 0, 3, 3, 8),
SpatialObject("Library", 1, 2, 0, 2, 2, 4)
]
complex_query = SpatialQuery(
question="Which public facilities are located in the northern part of the city and which are in the south?",
objects=city_objects,
context="City planning analysis"
)
complex_analysis = agent.analyze_spatial_query(complex_query)
print(f"\nQuery: {complex_query.question}")
print(f"\nAnalysis:\n{complex_analysis.answer}")
# Summary
print("\n" + "=" * 80)
print("SUMMARY")
print("=" * 80)
print("""
The Spatial Reasoning pattern enables agents to:
✓ Understand spatial relationships and positions
✓ Calculate distances and directions
✓ Find optimal paths and navigation routes
✓ Identify nearest neighbors
✓ Search within spatial regions
✓ Analyze complex spatial arrangements
✓ Combine geometric calculations with semantic understanding
This pattern is valuable for:
- Navigation and routing systems
- Layout and design optimization
- Geographic information systems
- Robotics and autonomous vehicles
- Game AI and virtual environments
- Spatial data analysis
- Map-based applications
""")
if __name__ == "__main__":
demonstrate_spatial_reasoning()
Agentic Design Pattern: Spatial Reasoning
Pattern 163: Spatial Reasoning – Line-by-Line Explanation
Overview
This document provides a detailed line-by-line explanation of the Spatial Reasoning pattern implementation.
Pattern Description
Pattern 163: Spatial Reasoning
Description: The Spatial Reasoning pattern enables agents to understand and reason about spatial relationships, positions, orientations, and geometric properties. This pattern helps agents solve problems involving physical space, layouts, navigation, and spatial configurations.
Components: 1. Coordinate System: Tracks positions and locations 2. Relationship Analyzer: Identifies spatial relationships (above, below, left, right, etc.) 3. Distance Calculator: Computes distances and proximity 4. Orientation Tracker: Manages directions and rotations 5. Spatial Query Engine: Answers spatial questions 6. Visualization Helper: Generates spatial descriptions
Use Cases: – Navigation and pathfinding – Layout design and optimization – Object placement and arrangement – Map understanding and interpretation – Geometry problem solving – Spatial puzzle solving – Floor plan analysis
Benefits: – Enables spatial understanding – Supports geometric reasoning – Facilitates navigation tasks – Improves layout optimization – Better scene understanding
Trade-offs: – Complex coordinate management – Requires clear spatial representation – May need visualization capabilities – Challenging for abstract spaces – Limited by 2D/3D representation
LangChain Implementation: Uses LLM for spatial relationship interpretation with structured coordinate systems and distance calculations. Combines symbolic representation with natural language understanding for spatial queries.
File Statistics
- Total Lines: 605
- Classes: 6
- Functions: 1
- Constants: 0
Module Documentation
"""
Pattern 163: Spatial Reasoning
Description:
The Spatial Reasoning pattern enables agents to understand and reason about spatial
relationships, positions, orientations, and geometric properties. This pattern helps
agents solve problems involving physical space, layouts, navigation, and spatial
configurations.
Components:
1. Coordinate System: Tracks positions and locations
2. Relationship Analyzer: Identifies spatial relationships (above, below, left, right, etc.)
3. Distance Calculator: Computes distances and proximity
4. Orientation Tracker: Manages directions and rotations
5. Spatial Query Engine: Answers spatial questions
6. Visualization Helper: Generates spatial descriptions
Use Cases:
- Navigation and pathfinding
- Layout design and optimization
- Object placement and arrangement
- Map understanding and interpretation
- Geometry problem solving
- Spatial puzzle solving
- Floor plan analysis
Benefits:
- Enables spatial understanding
- Supports geometric reasoning
- Facilitates navigation tasks
- Improves layout optimization
- Better scene understanding
Trade-offs:
- Complex coordinate management
- Requires clear spatial representation
- May need visualization capabilities
- Challenging for abstract spaces
- Limited by 2D/3D representation
LangChain Implementation:
Uses LLM for spatial relationship interpretation with structured coordinate
systems and distance calculations. Combines symbolic representation with
natural language understanding for spatial queries.
"""
Purpose: This module implements the Spatial Reasoning pattern, providing a structured approach to agent design and implementation.
Import Statements
Standard Library Imports
import dataclasses.dataclass
import enum.Enum
import langchain_core.output_parsers.StrOutputParser
import langchain_core.prompts.ChatPromptTemplate
import os
import typing.Any
import typing.Dict
import typing.List
import typing.Optional
import typing.Tuple
Purpose: Core Python functionality:
– dataclasses.dataclass: Simplified class definitions with @dataclass
– enum.Enum: Enumeration support
– langchain_core.output_parsers.StrOutputParser: Regular expression operations
– langchain_core.prompts.ChatPromptTemplate: Regular expression operations
– os: Environment variable access and system operations
– typing.Any: Type hints for better code documentation (List, Dict, Any, Optional)
– typing.Dict: Type hints for better code documentation (List, Dict, Any, Optional)
– typing.List: Type hints for better code documentation (List, Dict, Any, Optional)
– typing.Optional: Type hints for better code documentation (List, Dict, Any, Optional)
– typing.Tuple: Type hints for better code documentation (List, Dict, Any, Optional)
LangChain Imports
from langchain_openai.ChatOpenAI
Purpose: LangChain framework components: – LLM Interface: Wrapper for OpenAI models – Chains: Composable building blocks for complex workflows
Other Imports
from dotenv.load_dotenv
from math
Purpose: Additional dependencies for specialized functionality.
Classes
Class: Direction
Lines 61-72
Inherits from: Enum
Purpose:
Cardinal and ordinal directions
Class: SpatialRelation
Lines 75-88
Inherits from: Enum
Purpose:
Types of spatial relationships
Class: SpatialObject
Lines 92-101
Purpose:
Represents an object in space
Class: SpatialQuery
Lines 105-109
Purpose:
A spatial reasoning query
Class: SpatialAnalysis
Lines 113-119
Purpose:
Result of spatial reasoning
Class: SpatialReasoningAgent
Lines 122-416
Purpose:
Agent that performs spatial reasoning tasks
Methods:
__init__(self, model_name)
Lines 125-154
Purpose:
Initialize the spatial reasoning agent
Parameters:
– model_name: Language model instance
analyze_spatial_query(self, query)
Lines 156-182
Purpose:
Analyze a spatial query and return comprehensive analysis
Parameters:
– query: Input query or question
Returns: SpatialAnalysis
find_path(self, start, end, obstacles)
Lines 184-212
Purpose:
Find path from start to end avoiding obstacles
Parameters:
– start: Start
– end: End
– obstacles: Obstacles
Returns: <ast.Subscript object at 0x000001A06D17A650>
find_nearest(self, reference, candidates)
Lines 214-226
Purpose:
Find the nearest object to reference from candidates
Parameters:
– reference: Reference
– candidates: Candidates
Returns: <ast.Subscript object at 0x000001A06D60A550>
check_relationship(self, obj1, obj2, relation)
Lines 228-253
Purpose:
Check if a specific spatial relationship holds between two objects
Parameters:
– obj1: Obj1
– obj2: Obj2
– relation: Relation
Returns: bool
get_objects_in_region(self, objects, center, radius)
Lines 255-271
Purpose:
Get all objects within a spherical region
Parameters:
– objects: Objects
– center: Center
– radius: Radius
Returns: <ast.Subscript object at 0x000001A06D74A710>
_distance(self, obj1, obj2)
Lines 273-279
Purpose:
Calculate Euclidean distance between two objects
Parameters:
– obj1: Obj1
– obj2: Obj2
Returns: float
_calculate_distances(self, objects)
Lines 281-288
Purpose:
Calculate all pairwise distances
Parameters:
– objects: Objects
Returns: <ast.Subscript object at 0x000001A06D669750>
_identify_relationships(self, objects)
Lines 290-310
Purpose:
Identify spatial relationships between objects
Parameters:
– objects: Objects
Returns: <ast.Subscript object at 0x000001A06D63A550>
_get_direction(self, from_obj, to_obj)
Lines 312-340
Purpose:
Determine cardinal direction from one object to another
Parameters:
– from_obj: From obj
– to_obj: To obj
Returns: Direction
_check_obstacles(self, start, end, obstacles)
Lines 342-352
Purpose:
Check which obstacles are in the path between start and end
Parameters:
– start: Start
– end: End
– obstacles: Obstacles
Returns: <ast.Subscript object at 0x000001A06D7EFDD0>
_is_near_line(self, start, end, point, threshold)
Lines 354-374
Purpose:
Check if a point is near the line between start and end
Parameters:
– start: Start
– end: End
– point: Point
– threshold: Threshold
Returns: bool
_create_object_descriptions(self, objects)
Lines 376-384
Purpose:
Create textual descriptions of object positions
Parameters:
– objects: Objects
Returns: str
_create_visualization(self, objects)
Lines 386-416
Purpose:
Create ASCII visualization of object positions
Parameters:
– objects: Objects
Returns: str
Functions
demonstrate_spatial_reasoning()
Lines 419-600
Purpose:
Demonstrate spatial reasoning capabilities
Key Concepts Summary
Pattern Overview
The Spatial Reasoning pattern provides a structured approach to building AI agents with specific capabilities and behaviors. This implementation leverages LangChain’s composable components to create a robust and maintainable solution.
Core Components
- Agent Class: Main implementation of the pattern logic
- Prompt Engineering: Carefully crafted prompts to guide LLM behavior
- Tool Integration: Optional tools for extended functionality
- Memory Management: State handling across interactions
- Output Parsing: Structured response processing
When to Use
✅ Good for: – Scenarios requiring this specific pattern’s strengths – Applications needing structured agent behavior – Systems where explainability is important – Production environments requiring reliability
❌ Not ideal for: – Simple single-prompt tasks – Scenarios requiring different pattern characteristics – Resource-constrained environments (if computationally intensive)
Best Practices Demonstrated
1. Environment Configuration
- ✅ Use
.envfiles for API keys - ✅ Never hardcode credentials
- ✅ Validate environment variables on startup
2. Error Handling
- ✅ Graceful degradation on failures
- ✅ Informative error messages
- ✅ Logging for debugging
3. Prompt Engineering
- ✅ Clear, specific instructions
- ✅ Structured output formats
- ✅ Few-shot examples when beneficial
4. Code Organization
- ✅ Separation of concerns (class-based design)
- ✅ Type hints for clarity
- ✅ Docstrings for documentation
5. Performance Optimization
- ✅ Appropriate temperature settings
- ✅ Token budget management
- ✅ Caching where applicable
LangChain Components Used
- ChatOpenAI: OpenAI model interface
- Chains: Sequential component composition
Why LangChain? – Composable abstractions for LLM applications – Rich ecosystem of integrations – Production-ready patterns – Active community and development
Execution Flow
1. Initialize agent with configuration
↓
2. Load environment and dependencies
↓
3. Prepare prompt with input data
↓
4. Invoke LLM through LangChain
↓
5. Parse and validate response
↓
6. Return formatted result
Key Decision Points: – Model selection (temperature, model type) – Prompt structure (few-shot vs zero-shot) – Error handling strategy – Output format validation
Performance Considerations
Computational Cost
- LLM Calls: Primary cost factor (based on tokens)
- Latency: Depends on model and input size
- Memory: Minimal for most patterns
Optimization Strategies
- Prompt Optimization: Minimize tokens while maintaining clarity
- Caching: Cache repeated queries/responses
- Model Selection: Use smaller models when appropriate
- Batch Processing: Group similar requests
- Streaming: For long responses, stream tokens
Cost Estimates (Approximate)
- GPT-3.5-turbo: ~$0.002 per request
- GPT-4: ~$0.03 per request
- GPT-4-turbo: ~$0.01 per request
Actual costs vary based on input/output token counts
Common Pitfalls to Avoid
1. Prompt Design Issues
❌ Vague or ambiguous instructions
✅ Clear, specific, testable prompts
2. Error Handling
❌ Ignoring API errors or rate limits
✅ Implement retry logic and graceful degradation
3. Token Management
❌ Exceeding model token limits
✅ Truncate inputs, chunk processing
4. Security
❌ Exposing API keys in code
✅ Use environment variables
5. Testing
❌ Only testing happy paths
✅ Test edge cases, failures, malformed inputs
6. Cost Management
❌ Unmonitored API usage
✅ Implement budgets, alerts, and monitoring
Further Improvements
Production Enhancements
- Monitoring: Add logging, metrics, tracing
- Validation: Input/output schema validation
- Rate Limiting: Implement request throttling
- Caching: Redis or in-memory cache for responses
- Async Support: Non-blocking operations for scalability
Feature Extensions
- Multi-Model Support: Switch between providers
- Fallback Models: Graceful degradation strategy
- Custom Tools: Domain-specific function calling
- Advanced Memory: Vector stores for long-term memory
- Evaluation: Automated testing and benchmarking
Code Quality
- Testing: Unit tests, integration tests
- Documentation: API docs, usage examples
- Type Safety: Stricter type hints
- Configuration: Externalized config files
Related Patterns
- Pattern 162
- Pattern 164
Pattern Families: – Reasoning Patterns (001-011) – Multi-Agent Patterns (012-019) – Tool Use Patterns (020-025) – Memory Patterns (026-031) – And many more…
Summary
The Spatial Reasoning pattern provides a powerful approach to building AI agents with specific capabilities. This implementation demonstrates:
✅ Key Strengths: – Clean, maintainable code structure – LangChain integration for composability – Type hints and documentation – Production-ready patterns
🎯 Best Use Cases: – Applications requiring this pattern’s characteristics – Systems needing structured agent behavior – Production environments requiring reliability
📚 Learning Resources: – LangChain Documentation: https://python.langchain.com/ – Pattern Documentation: See related patterns – OpenAI API Reference: https://platform.openai.com/docs/
File: 163_spatial_reasoning.py
Total Lines: 605
Pattern Category: AI Agent Design Pattern #163
Generated from actual implementation code
For educational and reference purposes
Leave a Reply