BCI Exoskeleton Rehabilitation
BCI Exoskeleton Rehabilitation Arm for our Masters
🚀 [Project Overview]
Our final-year capstone project involved designing and building a wearable neuro-rehabilitation exoskeleton arm, controlled entirely via brain signals captured from a 14-sensor EEG headset (Emotiv Epoc X). The goal? Enable users—particularly stroke survivors—to initiate bicep curls and wrist rotations through mental imagery alone. This project page will lack images however a video is available, on youtube, I encourage you to watch. Github YouTube Video
[🧠 Core Vision]
We set out to prove that accessible, at-home neurotechnology could be used for real-world rehabilitation. Our prototype is:
Affordable: Uses consumer-grade parts (PLA, Arduino, Emotiv Epoc X)
Wearable & Portable: Shoulder-mounted, powered by batteries
Safe: Equipped with emergency stop systems, limit switches, and current monitoring
Modular: Designed for ease of future improvements or research
🔧 [What We Built]
Two servo-controlled joints: One for elbow flexion, one for wrist rotation
3D-printed PLA frame: Modular and adaptable for comfort and easy manufacturing
Safety features: Emergency stop, hardware limit switches, current sensors (INA226)
Distributed control: Three Arduino Unos and a Jetson Nano orchestrated via MQTT
Cloud-based EEG processing: Used Emotiv’s platform for brain signal interpretation
🔄 [Engineering Adaptations]
Despite initial ambitions of onboard ROS-based processing and carbon fiber materials, we adapted due to hardware, budget and time constraints:
Emotiv’s cloud processing replaced custom EEG decoding.
Control logic was distributed to Arduinos for stability via a Jetson Nano.
ROS and Gazebo simulations were dropped in favor of hands-on real-world testing.
PLA replaced carbon fiber to speed up and simplify prototyping.
👥 [Teamwork & Challenges]
We faced it all — burnouts, hardware faults, software crashes, and countless reprints — but we adapted, iterated, and supported each other. A few lessons we learned:
Training EEG models is highly individual and mentally taxing.
Simplicity wins under pressure — modular design saved us.
60+ tests validated the system across mechanical, electrical, software, and usability dimensions.
🧪 [Performance & Testing]
From power systems to user safety and fatigue, we validated each subsystem:
EEG tests confirmed signal stability and command accuracy.
Servo and safety tests proved reliable, smooth movement and safe halting.
Hardware tests included stress and bending analyses of printed parts.
User trials proved the system was wearable and responsive for extended use.
The only unfulfilled goal? Achieving 90° wrist rotation — we reached 45° due to time constraints.
🔮 [Future Possibilities]
This prototype lays the groundwork for exciting follow-up work:
Full 90° wrist articulation
Machine learning for adaptive signal recognition
Enhanced comfort and ergonomics
More nuanced mental command mapping for expanded motion
💬 [Final Reflections]
This wasn’t just an academic project — it was an engineering challenge, a team growth experience, and a vision for accessible neuro-rehabilitation. We built a robotic arm — but also built trust, resilience, and confidence in what’s possible with teamwork and bold thinking.
[📂 Resources]
📜 Final Technical Report PDF