5 AI-powered projects to automate your smart home

1. Automate air quality responses

❌ Basic setup: You check an air quality monitor manually or receive a notification about unhealthy levels.
✅ Smarter setup: Your home detects poor air quality and fixes it automatically.

How? Use CO₂, PM_2.5, or VOC sensors to monitor when values reach a threshold to trigger actuation that happens whether you’re at home or not.

• Turn on an exhaust fan when CO₂ is high.
• Send an alert and open a smart window for fresh air.
• Adjust your HVAC system for better circulation.

Follow this step-by-step codelab to automate your own air filtration system with air quality sensors using a Raspberry Pi, smart plug, and box fan.

2. Smarter lights with mmWave presence detection

❌ Basic setup: Use PIR motion sensors turn lights on/off based on movement which can lead to false positives, like pets or heating vents triggering a detection, and false negatives, like lights turning off when someone is sitting still. ✅ Smarter setup: Use millimeter wave (mmWave) sensors to detect human presence enabling precise, adaptive automation.

Use mmWave sensors to ensure smarter automation by detecting both static and moving presence, eliminating false triggers and improving energy efficiency. 

• Enable smarter lights that stay on when you're reading, working, or watching TV.
• Save energy by turning lights off when you leave a room.
• Monitor your sleep and detect falls.

Follow this step-by-step codelab to build your own presence detector with an mmwave sensor.

3. Enhance security with smarter detection using computer vision

❌ Basic setup: A traditional motion sensor triggers an alert for anything.
✅ Smarter setup: AI can distinguish pets, humans, and unknown visitors.

Use computer vision to differentiate between objects that you actually care about.

• Ignore pets while detecting humans.
• Trigger an alarm only for unknown faces.
• Turn on outdoor lights and record footage when unusual visitors are detected.

Follow this step-by-step codelab to set up people detection notifications in Home Assistant using Computer Vision.

‍

4. Train your own custom models for easy snacking

❌ Basic setup: You’re browsing open-source registries for existing models that kind of do what you want them to do.
✅ Smarter setup: Train your own AI models for improved accuracy and customization.

How?

• Refill a snack bowl when it’s empty.
• Dispense treats for humans or pets when a certain pose is detected.
• Log patterns in snack consumption to predict when inventory should be replenished.

Follow this step-by-step codelab to build a smart snack dispenser with CV.

5. Optimize plant care with a local LLM and machine learning

❌ Basic setup: A water pump that waters a plant based only on soil moisture levels.
✅ Smarter setup: Use speech-to-text and a local LLM to get personalized plant care recommendations.

Instead of a one-size-fits-all approach, AI-powered plant care adapts to each plant’s unique needs, optimizing growth and reducing waste. 

• Identify plants automatically with an AI model to ensure species-specific care.
• Track soil conditions in real time to detect early signs of overwatering, dehydration, or nutrient deficiencies.
• Analyze water usage patterns to predict potential environmental issues and adjust irrigation accordingly. 

Follow this step-by-step tutorial to use a local LLM and machine learning for plant care.

Take your smart home to the next level

Modern smart homes should go beyond simple, pre-programmed commands. They can be multi-modal, integrating input from various sensors and seamlessly triggering multiple smart devices to anticipate your needs rather than just react. 

When designing your home automation setup, there are countless possibilities, from cloud-based, local, or hybrid approaches. Choose what best fits your needs.

Try one of these five upgrades and start making your smart home actually smart. Got a cool automation setup? Share it with us in the Viam Discord server in the #built-on-viam channel.

‍