Roadmap

Vision: ESPectre aims to democratize Wi-Fi sensing by providing an open-source, privacy-first motion detection system with a path toward machine learning-powered gesture recognition, Human Activity Recognition (HAR), and 3D indoor localization.

This roadmap outlines the evolution from the current mathematical approach (just IDLE/MOTION) toward ML-enhanced capabilities (Gesture detection, Human Activity Recognition) and advanced spatial sensing (3D localization via phase-coherent multi-node arrays), while maintaining the project's core principles: community-friendly, vendor-neutral, and privacy-first.

Table of Contents

Market Opportunity

The global Wi-Fi sensing market is experiencing rapid growth, driven by demand for non-intrusive, privacy-preserving sensing solutions.

Metric Value Source
Market Size (2024) $2.1B Allied Market Research
Projected Size (2030) $12.5B Allied Market Research
CAGR 34.2% 2024-2030

Key Drivers

  • Privacy concerns: Camera-free sensing for elderly care, healthcare, and smart homes
  • Cost efficiency: Leverages existing WiFi infrastructure (no additional hardware)
  • Regulatory push: IEEE 802.11bf (Wi-Fi Sensing) standardization in progress

Target Applications

Application Market Segment ESPectre Capability
Smart Home Consumer IoT Motion detection, presence sensing
Elderly Care Healthcare Fall detection, activity monitoring
Security Commercial Intrusion detection, occupancy
Retail Analytics Enterprise People counting, traffic flow
Gesture Control Consumer Electronics Hands-free device interaction
Indoor Localization Logistics/Retail Asset tracking, navigation (30-50cm accuracy)

Competitive Positioning

Competitor Approach ESPectre Advantage
Origin Wireless Proprietary, cloud-dependent Open-source, edge-first, no subscription
Cognitive Systems Enterprise-only, high cost Affordable ($5 hardware), DIY-friendly

ESPectre is uniquely positioned as the only open-source, production-ready WiFi sensing platform with native smart home integration.

Current State

ESPectre v2.x provides a motion detection system using mathematical algorithms:

Component Status Description
MVS Algorithm Production Moving Variance Segmentation for motion detection
Band Calibration Production Automatic subcarrier selection (NBVI)
ESPHome Integration Production Native Home Assistant integration with auto-discovery
Micro-ESPectre Production Python R&D platform for rapid prototyping
ML Data Collection Ready Infrastructure for labeled CSI dataset creation
Analysis Tools Ready Comprehensive suite for CSI analysis and validation

Timeline Overview

     Q1 2026              Q2-Q3 2026              Q4 2026 - Q4 2027
        │                     │                          │
        ▼                     ▼                          ▼
┌───────────────┐     ┌───────────────┐     ┌─────────────────────┐
│  SHORT-TERM   │────▶│   MID-TERM    │────▶│     LONG-TERM       │
│   3-6 months  │     │   6-12 months │     │    12-24 months     │
├───────────────┤     ├───────────────┤     ├─────────────────────┤
│ Data & Docs   │     │ ML Models     │     │ 3D Localization     │
│ Dataset infra │     │ Training      │     │ Advanced Apps       │
│ Tooling       │     │ Edge Inference│     │ Multi-sensor Fusion │
└───────────────┘     └───────────────┘     └─────────────────────┘

Short-Term (3-6 months)

Focus: Data collection, documentation, and ML groundwork.

Data & Datasets

Task Priority Status
Expand labeled CSI dataset (gestures, activities) High Planned
Community data contribution guidelines High Planned
Dataset versioning and reproducibility Medium Planned
Multi-environment data collection (offices, homes, industrial) Medium Planned

Documentation & Tooling

Task Priority Status
Feature extraction pipeline documentation High Planned
Data labeling best practices guide Medium Planned
Jupyter notebooks for CSI exploration Medium Planned
Automated data quality validation Low Planned

Infrastructure

Task Priority Status
Standardized dataset format (HDF5 or extended NPZ) Medium Planned
Dataset registry and metadata management Low Planned

Mid-Term (6-12 months)

Focus: ML model development, training infrastructure, and initial inference capabilities.

Model Development

Task Priority Status
Gesture recognition models (RF, CNN, LSTM) High Planned
Human Activity Recognition (HAR) models High Planned
People counting / presence estimation Medium Planned
Fall detection Medium Planned

Training Infrastructure

Task Priority Status
Centralized training experiments (local) High Planned
Model versioning and experiment tracking High Planned
Hyperparameter optimization pipelines Medium Planned
Cross-validation with diverse environments Medium Planned

Inference

Task Priority Status
Edge inference on ESP32 (manual MLP) High Done
TensorFlow Lite Micro integration Medium Exploratory
Model optimization (quantization, pruning) Medium Exploratory
Latency and memory profiling Medium Planned

Long-Term (12-24 months)

Focus: 3D indoor localization and advanced applications.

3D Localization

Goal: Transform motion detection into real-time 3D indoor localization with 30-50 cm accuracy.

This capability represents a significant leap from binary motion detection to precise spatial tracking, enabling applications like indoor navigation, asset tracking, and advanced gesture recognition.

Capability Description
Technology Phase-coherent multi-node array (3-4× ESP32-C5)
Frequency 5GHz WiFi 6 for improved accuracy
Algorithm MUSIC (Multiple Signal Classification) for AoA triangulation
Target Accuracy 30-50 cm in 3D space
Hardware Cost Stage-gated: devkit cluster first, custom hardware later
Task Priority Status
Wired shared-clock phase coherence validation (2-device prototype) High Research
AoA estimation proof-of-concept High Research
Wireless clock discipline + ping-pong reference calibration prototype High Research
Architecture trade-off study (wired shared-clock vs wireless disciplined sync) High Research
Decision gate: select long-term architecture from benchmark results High Research
Node count scaling policy (3 -> 4 based on RMS/availability) Medium Research
Custom carrier/backplane (optional, post-validation) Medium Research
MUSIC algorithm implementation Medium Research
5GHz CSI extraction validation Medium Research

Advanced Applications

Task Priority Status
Multi-sensor fusion (multiple ESP32 devices) Medium Exploratory
Room-level activity tracking Medium Exploratory
Vital signs monitoring (breathing, heartbeat) Low Research
Integration with IEEE 802.11bf (Wi-Fi Sensing standard) Low Research

Architecture Evolution

ESPectre's architecture evolves through three major versions, each adding capabilities while maintaining backward compatibility.

┌─────────────────────────────────────────────────────────────────────────────┐
│                        ARCHITECTURE EVOLUTION                               │
├─────────────────────────────────────────────────────────────────────────────┤
│                                                                             │
│  v2.x (Current)          v3.x (ML-Enhanced)         v4.x (3D Spatial)       │
│  ───────────────         ─────────────────         ────────────────         │
│                                                                             │
│  ┌───────────┐           ┌───────────┐             ┌───────────────┐        │
│  │  ESP32    │           │  ESP32    │             │ 4× ESP32-C5   │        │
│  │  ┌─────┐  │           │  ┌─────┐  │             │ Phase-Coherent│        │
│  │  │ CSI │  │           │  │ CSI │  │             │   ┌─────┐     │        │
│  │  └──┬──┘  │           │  └──┬──┘  │             │   │ CSI │     │        │
│  │     │     │           │     │     │             │   └──┬──┘     │        │
│  │  ┌──▼──┐  │           │  ┌──▼──┐  │             └──────┼────────┘        │
│  │  │ MVS │  │           │  │ MVS │  │                    │                 │
│  │  └──┬──┘  │           │  └──┬──┘  │             ┌──────▼────────┐        │
│  └─────┼─────┘           │  ┌──▼──┐  │             │  Local/Cloud  │        │
│        │                 │  │ ML  │  │             │  ┌─────────┐  │        │
│        │                 │  │Edge │  │             │  │ MUSIC   │  │        │
│        │                 │  └──┬──┘  │             │  │Algorithm│  │        │
│        │                 └─────┼─────┘             │  └────┬────┘  │        │
│        │                       │                   │  ┌────▼────┐  │        │
│        │                       │                   │  │ 3D Pos  │  │        │
│        ▼                       ▼                   │  │ (X,Y,Z) │  │        │
│  ┌──────────┐            ┌──────────┐              │  └────┬────┘  │        │
│  │   Home   │            │   Home   │              └───────┼───────┘        │
│  │Assistant │            │Assistant │                      │                │
│  └──────────┘            └──────────┘                      ▼                │
│                                                      ┌──────────┐           │
│  Output:                 Output:                     │   Home   │           │
│  IDLE/MOTION             Gesture, HAR,               │Assistant │           │
│                          Fall Detection              └──────────┘           │
│                                                                             │
│                                                      Output:                │
│                                                      3D Position            │
│                                                                             │
└─────────────────────────────────────────────────────────────────────────────┘
Version Capability Processing Key Innovation
v2.x Motion detection (IDLE/MOTION) 100% Edge MVS algorithm, auto-calibration
v3.x Gesture, HAR, fall detection 100% Edge TFLite Micro inference
v4.x 3D indoor localization Edge + Local/Cloud Phase-coherent multi-node array

Principles & Governance

ESPectre is committed to open-source principles and community-driven development.

Core Principles

Principle Description
Edge-First All processing happens locally on ESP32 - no cloud dependency
Privacy-Preserving CSI data never leaves the device; no cameras, no recordings
Hardware-Agnostic Supports ESP32, ESP32-S2/S3, ESP32-C3/C5/C6 variants
Open Development All development happens in the open on GitHub
Reproducibility Experiments and results must be reproducible

Governance

Aspect Approach
License GPLv3 - ensures software remains free and open source
Decision Making Maintainer-led with community input via GitHub Discussions
Roadmap Updates Quarterly reviews based on community feedback and resources

Contributing

We welcome contributions! See CONTRIBUTING.md for: - Code contribution guidelines - Data contribution guidelines - Development setup - Code style and commit conventions

How to Propose Changes

This roadmap evolves with community input. Here's how you can contribute:

Method Use Case
GitHub Issues Propose new features or report blockers for existing items
GitHub Discussions Discuss priorities, trade-offs, and architectural decisions
Pull Request Submit changes to this file with your proposal

Process

  1. Check existing items - Review current roadmap and open issues
  2. Open an Issue - Describe your proposal with use case and rationale
  3. Discuss - Engage with maintainers and community in the issue/discussion
  4. Submit PR - Once there's consensus, update this file via Pull Request

Roadmap Updates

This roadmap is reviewed and updated quarterly. Last update: February 2026

For the latest status and discussion: - GitHub Issues - GitHub Discussions

License

GPLv3 - See LICENSE for details.