Home Automation Systems

🧠 How I put a brain in my house ... or how I went from basement Raspberry Pis to a fully orchestrated smart home 🏡🔌. What started as a small experiment with a Raspberry Pi 1 running a few bash scripts has quietly evolved into a fully integrated, self-hosted smart home ecosystem — and I thought it was time to share a glimpse into it. At the heart of this setup lies a Raspberry Pi 2 running Home Assistant, acting as the brain of the operation. It's not just a dashboard — it's the central hub that orchestrates a complex symphony of automation, energy management, and remote control. 📡 Remote Access via WireGuard VPN Secure access is a must. Whether I’m on vacation or just checking in from the office, a WireGuard VPN ensures encrypted access to my network and devices. There is no other way to get into my local network. 📺 Media & Messaging A Raspberry Pi 4 runs Plex, providing local media streaming to all devices. A Raspberry Pi 3 handles MQTT, the message bus connecting many of the real-time devices, such as my custom devices mostly running on small ESP32 units. 💡 Automation & Environment Netatmo sensors (indoor/outdoor + rain gauge) track environmental conditions. The Philips Hue Bridge controls all Zigbee-connected lighting. Somfy TaHoma takes care of most of the roller shutters and blinds with real-time outdoor sensor data. 🌡️ The Power of KNX In the core of my home I've placed KNX to the mix — a robust, bus-based system connecting wall switches, temperature sensors, and intelligent floor heating throughout the house. KNX ties into both the electric cabinet and Home Assistant (via an IP control module), creating a seamless control layer that operates even without cloud connectivity. 🔌 Energy & Power Flow A small gadget called Wattwächter is placed on top of the electric meter to monitor real-time consumption. In addition, the EcoFlow Stream and my wallbox (managed via OPCC) support energy storage and EV charging, with integration into the broader Home Assistant logic. 🌿 Garden Control via ESP32 ESP32 boards manage two unique zones: - Monitoring the water level in the garden reservoir - Controlling a DIY robotic lawn mower connection - as unfortunately my mower only supports an ad-hoc Bluetooth connection. 💻 And yes — Raspberry Pi 1 is still hanging in there, running lightweight custom scripts with cronjobs like it’s 2014. Just with some intelligence... It grabs the state from Home Assistant via its RESTful API - then gives this state to an LLM which can make suggestions. These are then send to me or (if whitelisted) immediately executed. This project taught me more than just YAML and MQTT topics — it reshaped how I think about distributed systems, resilience, and user-centric design at home. #SmartHome #HomeAssistant #IoT #Automation #AI

I have ten smart home apps on my phone. Ten different ecosystems with zero shared control plane. So I replaced them all with a single Docker container on a $50 Raspberry Pi. Home Assistant is open source, runs locally, supports 2,000+ integrations. No subscriptions, no vendor lock-in. My data stays on my hardware. But the real payoff was building a Python CLI on top of it: → hactl thermostat set-temp 72 to adjust temperature → hactl mode bedtime to lock doors, dim lights, lower thermostat → hactl vacuum start to clean the house → hactl notify say "Dinner is ready" for voice announcements 13 command groups wrapping HA's REST API with Typer + httpx. Why a CLI? Because Claude Code can call it. I type "bump the temp down" in my coding terminal and the house responds. No app, no phone, no voice command. It goes further. I run an autonomous agent framework where persistent AI agents operate on their own schedules. One agent checks the weather forecast and adjusts the thermostat before I'm at my desk. Another runs hactl mode bedtime at 11 PM if I'm still working. The house becomes part of the agent's toolkit. AI-assisted home control, both interactive and autonomous, from the same place I write code. Full technical guide with code examples in first comment. #HomeAssistant #SmartHome #OpenSource

Smart home meets control theory. Energy savings without compromising comfort. In a previous post, I have shown a method for manipulating a controller by superimposing a simple control loop based on a dynamic setpoint. (https://lnkd.in/e4A8HZ4y) On the following slides I will show you how to bring the control theory into practice. Slide 1: Control theory can make smart home devices even smarter. In my private office room I have a smart heating system as a part of my home energy management system (#hems). The radiator thermostat from tado° has only an input value for the setpoint temperature and an output value for the current temperature. You can also set schedules and recognise whether you are at home or on the move. To extend the functionality with an external temperature sensor and another sensor for room occupancy, you need a management system such as Home Assistant. As the internal temperature sensor is located close to the heating source, it is better to use an external sensor placed in the centre of the room. To detect whether a person is present, a mmWave radar sensor recognises micro-movements much better than passive infrared (PIR) sensors. Both sensors are coupled with #zigbee and can be purchased inexpensively from SONOFF. To indirectly influence the valve position of the thermostat, it makes sense to increase the setpoint temperature depending on the actual value in order to improve the dynamics due to a proportional factor K. If a person leaves the room for longer than a certain time, the setpoint temperature is lowered to save energy. Slide 2: Include other influencing variables in order to set the correct setpoint temperature. I also added an IKEA door and window reed contact over zigbee to pause heating mode. To extend the logic of chosing the right setpoint temperature, the time can also take into account. W1 is the setpoint when a person is present, W2-W4 are different setpoints for short absences, longer breaks or night mode. The greater the difference between the setpoint temperature W and the current temperature Y, the greater the amplified setpoint W*. Heating radiators have a thermal mass, which leads to a heat capacity. It therefore takes a certain amount of time for the temperature to be reached. In addition, the stored energy heats the room further afterwards. You can learn more about this in a previous article on heat transfer system characteristics. (https://lnkd.in/e-ZJeNJk) Conclusion: Smart home devices such as radiator thermostats have their own control strategy. For reasons of cost efficiency, they use integrated temperature sensors to regulate the estimated room temperature to a fixed setpoint. The use of external sensors may increase comfort. Presence sensors can be included to reliably detect the actual occupancy of the room and not the surroundings of the building. They can also be used to automate the room light. But that might be another story. #engineering #controlsystems #controltheory #smarthome

[Nope. This is not another post about AI agents ;)] The "Smart Home" trap is real. We’ve all been sold the promise of a seamless ecosystem where our solar, batteries, and vehicles work in perfect harmony. But the reality? We’re living in a world of vendor silos. My Solar Batteries, Microinverters, Electrical Panels, detachable batteries, and EV charger are all "smart." Yet, individually, they have no idea what the others are doing. They each operate on their own cloud, their own assumptions, and their own limitations. I decided to stop waiting for vendors to play nice and instead built my own orchestration layer and control plane. I’ve spent the last week treating my home energy setup not like a collection of appliances, but as a distributed control system. Using Home Assistant as a substrate, I’ve built a custom "Home Energy OS" that: - Fuses multi-vendor telemetry: Moving beyond proprietary app data to a single, queryable state graph. - Enforces a "Policy Contract": Before writing code, I defined a plain-English policy document that dictates what the house is forbidden from doing—ensuring safety and efficiency above all else. - Replaces guessing with verification: Using SPAN’s circuit-level metering to verify that an action actually had a physical effect, rather than just trusting a "success" flag from an API. The goal wasn't just to monitor energy; it was to manage it with the same rigor we use for enterprise infrastructure. The interesting capabilities are emergent from the fusion, not present in any single product. Whether you're an architect, an engineer, or just someone who wants their home to actually think for itself, you might find the architectural breakdown useful. Full technical details—including how I handled per-array solar forecasting and the data-fusion logic in the comments: I’d love to hear how you’re bridging the gaps in your own home infrastructure. #EnergyTech #SystemsEngineering #HomeAssistant #Infrastructure #Automation #Architecture #SmartHome