Overview

​

What is AveniECA?

​

The ECA Model

​

Episodic

​

Cybernetic

​

Agency

​

Core Concepts

​

Digital Twins

• Has a unique module_id that identifies its type (e.g., “air_conditioner”, “temperature”)
• Maintains its own state history through ESS (Episodic State Space)
• Can be streamed to via Kafka using sub_topics
• Can exist as individual modules or be combined into aggregates

​

Signals

• state: A vector of float values representing the twin’s current state
• valence: An optional emotional/evaluative assessment of the state
• score: An optional numeric ranking or importance value
• emb_inp: An optional reference to embedding inputs for semantic understanding

​

Modules

• Group related twins together (e.g., all “temperature” sensors)
• Can represent individual entities or aggregate combinations
• Store their state history in the Episodic State Space (ESS)
• Enable predictions through the Cortex API

​

Sequences

• Link multiple ESS entries together to form temporal patterns
• Enable the system to learn from past state transitions
• Support prediction of future states based on historical sequences
• Can have different statuses to categorize types of experiences

​

How the System Works

​

Data Flow

1. Capture: Real-world state is captured and converted into a Signal
2. Stream: Signals are published to Kafka topics associated with digital twins
3. Store: The AveniECA instance stores signals as ESS entries with module_id
4. Sequence: ESS entries are organized into sequences representing episodes
5. Learn: The system identifies patterns in sequences of state transitions
6. Predict: The Cortex API predicts likely next states based on learned patterns

​

Example Workflow

from avenieca.data import Signal
from avenieca.producers import Stream
from avenieca.config.broker import Broker
import os

# Define how to capture state
def capture_temperature():
    # In real application, read from sensor
    temp = 22.5
    return Signal(
        state=[temp],
        valence=10.0 if temp < 25 else -10.0,
        score=1
    )

# Configure connection to digital twin
config = Broker(
    url=os.environ["KAFKA_URL"],
    sub_topic="temperature_sensor_1",  # Twin's subscriber topic
    group="sensors",
    pub_topic=""
)

# Stream continuous state updates
stream = Stream(config=config, sync_rate=1.0)
stream.publish(capture_temperature)

1. Captures temperature readings as Signals
2. Publishes them to the “temperature_sensor_1” twin
3. Builds up episodic memory in the ESS
4. Enables prediction of future temperature states

​

Aggregates: Combining Multiple Twins

# Individual modules
temperature = ESSResponse(module_id="temperature", state=[28.0], valence=-90)
humidity = ESSResponse(module_id="humidity", state=[65.0], valence=-50)
air_quality = ESSResponse(module_id="air_quality", state=[85.0], valence=-70)

# Aggregate combines them
aggregate = ESSInsert(
    module_id="climate_aggregate",
    state=[28.0, 65.0, 85.0],  # Combined states
    aggregate_module_id=["temperature", "humidity", "air_quality"],
    aggregate_shape=[1, 1, 1],  # Size of each component
    # ... other aggregate metadata
)

• Multi-modal predictions: Predict how multiple related systems evolve together
• Holistic understanding: Capture relationships between different aspects
• Efficient querying: Get complete system state in one operation

​

Next Steps