Reproduce “Trail: Accurate Local Smart Home Control Using Decision Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Maleeha Masood, Deepak Vasisht, Indranil Gupta
Trail uses a local model to choose rooms, devices, and operations from a home’s runtime registry rather than generate complete commands. It makes sequential typed decisions and refuses when path confidence falls below a threshold. On simulated Trek-Robust, exact-match accuracy is 85.3% versus 69.8% for finetuned Qwen3.5-4B, but Trail trails on Trek-Base; neither benchmark establishes deployment safety.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.