LangChain_digital_auto

Setup Guide to set up a LangChain/LangGraph-based multi-agent system that uses a locally running Ollama LLM to interpret vehicle sensor data like rain intensity or hood status from the Kuksa data broker and autonomously sets the appropriate windshield wiper mode via VSS actuators.

Hierarchy

 ├── SmartWiperAgents <- local Langgraph agents
 ├── SmartWiperApp <- Velocitas Runtime 

Requirements

Tool	Version	Purpose
Ubuntu / WSL2	24.04+	Host OS
Python	3.12+	LangGraph runtime
Docker	24+	Kuksa databroker container
LangSmith API key	—	Studio tracing (smith.langchain.com)

Velocitas Runtime Setup (SmartWiperApp)

this creates the Velocitas-Runtime (Kuksa Databroker, MQTT, Mock-Service) that is used by the "vehicle" model inside "SmartWiperAgents"

1. Clone the template repo

git clone https://github.com/eclipse-velocitas/vehicle-app-python-template.git SmartWiperApp
cd SmartWiperApp

2. pull the packages declared in .velocitas.json

velocitas init

3. Sync devcontainer / scripts / workflows

velocitas sync

4. Start Velocitas Runtime

velocitas exec runtime-local up

SmartWiperAgents Setup

SmartWiperAgents is a multi-agent system using LangGraph that responds to VSS data from the Velocitas Vehicle Runtime and decides on the windshield wiper mode based on the input.

1. Requirement checks

python3.12 --version

if not installed:

 sudo apt install python3.12

2. Local Ollama installation

install ubuntu binary and set up a system service so it can run in the background:

curl -fsSL https://ollama.com/install.sh | sh

start it with

ollama serve

pull the Ollama 3.1:8b model

ollama pull llama3.1:8b

3. Setup virtual environment

python3.12 -m venv .venv

to activate the environment for the rest of the setup

source .venv/bin/activate

4. install requirements

pip install -r requirements.txt

5. add init.py

to define the directories as packages and allow import:

touch app/__init__.py
touch agents/__init__.py
touch graph/__init__.py

6. export vehicle module

in the virtual environment

export PYTHONPATH="/your/path/to/the/project/SmartWiperApp/gen/vehicle_model:$PYTHONPATH"

7. start SmartWiper Vehicle App

to start the Velocitas-Vehicle-App run from /SmartWiperAgents

python -m app.SmartWiper

KUKSA client setup

to change the standart values of the Velociats Runtime while is is running, install the KUKSA client:

pip install kuksa-client

start the KUKSA client:

kuksa-client grpc://127.0.0.1:55555

to change the standart values use this:

setValue Vehicle.Speed 0
setValue Vehicle.Body.Windshield.Front.Wiping.Mode "MEDIUM"
setValue Vehicle.Body.Hood.IsOpen true

Logging with LangGraph Studio Setup

Prerequisite: Have a account on smith.langchain.com

1. install LangGraph CLI

pip install -U "langgraph-cli[inmem]"

2. If Ollama is not running, load and run Ollama

ollama pull llama3.1:8b
ollama serve

3. Setup Environment Variables ``.env``` in project root

# LangSmith Tracing
LANGSMITH_API_KEY=lsv2_pt_xxxxxxxxxxxxxxxxxxxx #from Langsmith website
LANGSMITH_TRACING=true
LANGSMITH_PROJECT=SmartWiperAgents
LANGSMITH_ENDPOINT=https://api.smith.langchain.eu #must match the website where the Langsmith account is registered

# Ollama
OLLAMA_MODEL=llama3.1:8b

if outside of the EU, use LANGSMITH_ENDPOINT=https://api.smith.langchain.com The API key can be generate under settings -> API keys

4. configure langgraph.json in project root

{
  "$schema": "https://langgra.ph/schema.json",
  "dependencies": ["."],
  "graphs": {
    "wiper_graph": "./graph/wiper_graph.py:graph"
  },
  "env": ".env",
  "python_version": "3.12"
}

5. install dependencies

if not already installed in requirements.txt

pip install -U \
  "langgraph-cli[inmem]" \
  langgraph \
  langchain-core \
  langchain-ollama \
  langsmith \
  python-dotenv

6. start LangGraph Studio

langgraph dev

Setup benchmarking

to use the benchmarking, psutil needs to be installed

pip install psutil