Locentra 室內定位研究與模擬平台

Locentra: Making the Invisible World of Indoor Positioning Visible 👻📡

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine

What I Built

Locentra is an interactive indoor-localisation research and simulation platform designed to explore how wireless positioning systems estimate the location of people and objects inside buildings.

Inspired by the theme Ghost in the Machine, Locentra explores the invisible signals and mathematical intelligence operating behind indoor positioning.

GPS works well outdoors, but indoor environments introduce challenges such as signal reflections, obstacles, packet loss, and inaccurate distance measurements.

Locentra provides a virtual laboratory where these challenges can be simulated, visualised, and evaluated.

Key Features

Interactive 2D & 3D Spatial Lab: Visualise indoor rooms, positioning anchors, and target trajectories.
Wireless Measurement Simulation: Simulate Gaussian noise, packet dropouts, NLOS bias, and measurement outliers.
Multiple Positioning Algorithms: Compare Non-Linear Least Squares, Robust Huber Estimation, and Kalman Filtering.
Benchmark Arena: Evaluate algorithms using RMSE, MAE, P95 error, and valid-estimate rates.
Scenario Studio: Configure room geometry, anchor arrangements, and simulation parameters.
Dataset Explorer: Explore and analyse generated positioning datasets.

The idea is to make invisible wireless measurements visible, understandable, and measurable.

Demo

🚀 Live Application: https://locentra-olive.vercel.app/

⚙️ Live Backend API: https://locentra.onrender.com/

Code

💻 GitHub Repository: https://github.com/badgujarkunal93-blip/Locentra

The project is open source, and contributions, suggestions, and improvements are welcome!

How I Built It

Locentra combines a modern web interface with a Python-based numerical positioning engine.

Frontend

React 19
TypeScript
Vite
Tailwind CSS
Three.js
React Three Fiber
React Three Drei
Zustand
TanStack React Query
Recharts

Backend

Python
FastAPI
NumPy
SciPy
Pandas
Pydantic
Uvicorn

Deployment

Vercel: Frontend hosting and deployment.
Render: Backend hosting and FastAPI deployment.

How the System Works

Virtual Room Configuration → Anchor Placement → Synthetic Measurement Generation → Position Estimation → Algorithm Comparison → Performance Visualisation

The numerical engine uses mathematical optimisation and filtering techniques to estimate target positions from simulated wireless distance measurements.

Robust estimation techniques help reduce the impact of noisy and corrupted measurements, while the interactive interface makes algorithm behaviour easier to understand.

Why Does Open Innovation Matter?

Open innovation matters because advanced positioning technology should not remain limited to expensive hardware laboratories and proprietary systems.

By building Locentra using open-source technologies such as NumPy, SciPy, FastAPI, React, and Three.js, the project makes experimentation and algorithm evaluation more accessible.

Developers and researchers can inspect the implementation, experiment with different positioning algorithms, modify simulation parameters, and extend the platform.

It also gives the community an opportunity to explore indoor localisation before investing in physical positioning hardware.

The goal is to make research more transparent, reproducible, and accessible.

My Agent Session

Locentra was developed through an AI-assisted workflow involving architecture planning, implementation, debugging, testing, and iterative refinement.

The development process involved validating the numerical positioning engine, improving the simulation pipeline, and refining the interactive visualisation experience.

The Spirit of It

Imagine entering a building where GPS stops working.

Your device still exists. The room still exists. But locating your exact position becomes much harder.

Behind the scenes, invisible wireless signals travel between anchors and devices. Locentra recreates that invisible world through mathematics, simulations, and interactive visualisation.

The ghost is not the device. It is the invisible intelligence working behind the signals.

Locentra is my attempt to make that invisible intelligence visible.

Prize Categories

Overall Challenge

Hacktoberfest Open-Source AI Challenge — Week 4: Ghost in the Machine

Partner Categories

Best Use of Render ($200) — Locentra uses Render to deploy and host its Python FastAPI backend. The service exposes health-check and simulation endpoints, connecting the frontend to the numerical positioning engine through HTTP APIs.
Best Use of GitHub Copilot ($100) — AI-assisted development of the Locentra codebase, including implementation, debugging, and refinement. (Include this category only if GitHub Copilot was actually used.)