Equilivest

A wearable robotic vest that delivers vibrotactile biofeedback to post-stroke patients, helping them recover dynamic walking balance through iterative neurorehabilitation protocols.

Project Overview

Equilivest is a smart wearable device that reads inertial data from IMU sensors embedded in a vest and delivers timely vibrotactile stimulation to guide the wearer’s postural correction. The system was developed with stroke survivors in mind — a population that frequently retains chronic balance deficits after discharge from acute care.

Three therapeutic approaches are under investigation: Artificial Vestibular Feedback, Gait Pacemaker, and Risk-Predictor (fall anticipation). The project integrates signal processing, embedded firmware, EEG-based intention detection, and clinical validation.

Scope

The project is structured in three successive stages:

1. Stage 1 — Design and fabricate the vest; embed IMU sensors and vibrotactile actuators; test the core clinical hypothesis on dynamic balance feedback.
2. Stage 2 — Build a reliable fall-risk predictor from inertial and gait signals; integrate real-time risk assessment into the feedback loop.
3. Stage 3 — Read EEG signals to detect the patient’s motor intention before movement initiation, closing the loop between neural intent and assistive stimulation.

Milestones

1. Concept & clinical framing — Patient surveys and caregiver interviews shaped three therapeutic strategies: Artificial Vestibular Feedback, Gait Pacemaker, and Risk-Predictor.
2. First vest prototype — IMU sensing + vibrotactile feedback hardware integrated into a wearable garment; initial bench tests completed.
3. IROS 2022 Workshop presentation — Work presented at the Assistive Robotic Systems for Human Balancing and Walking workshop at IROS 2022.
4. arXiv publication (January 2023) — Full system description and methodology published as arXiv:2301.06528.
5. Clinical hypothesis testing — Early-stage patient trials to validate the vestibular feedback concept in a rehabilitation setting.
6. Fall predictor development (ongoing) — Machine learning pipeline on IMU/gait signals for real-time fall-risk estimation.
7. EEG integration (ongoing) — BCI layer to decode movement intention from cortical signals and trigger anticipatory feedback.
8. Undergraduate & graduate theses (2023–2025) — Three academic theses completed, covering hardware, firmware, and signal processing aspects.

Contributions

• End-to-end design of the wearable sensing and actuation hardware
• Real-time vibrotactile feedback firmware for embedded targets
• Signal processing pipelines for IMU-based gait and balance analysis
• Clinical protocol design in collaboration with rehabilitation specialists
• Fall-risk prediction models from wearable sensor data
• EEG-based motor intention decoding (BCI layer, Stage 3)
• Peer-reviewed publication at IROS 2022 workshop; arXiv preprint

Collaborators

• Eng. Nicolás Vargas Alice
• Eng. Marina Cristina Perez Gaido
• Tasos Papastylonou — University of Essex, UK

Videos

Publications & Theses

Conference / Preprint

• Equilivest: A Robotic Vest to aid in Post-Stroke Dynamic Balance Rehabilitation — arXiv:2301.06528

Theses

• BS Thesis 1 — Equilivest: A robotic vest to aid in post-stroke dynamic balance rehabilitation
• BS Thesis 2 — Equilivest: Investigación de modelos de inteligencia artificial para la predicción de caídas
• MS / Graduate Thesis — Predicting Falls using Time Series Data from the Equilivest Device