Bionikle
A low-cost, high-feature bionic ankle.

Bionikle is an affordable powered bionic ankle built on 3D printing, on-device AI, and custom real-time control, bringing capable, adaptive mobility within reach.

Mobility, re-engineered from the ground up

Bionikle is a powered bionic ankle that adapts to how you move. Onboard sensing reads your gait in real time, an AI classifier recognises what you're doing (walking, climbing stairs, descending) and a custom controller drives the joint with the right motion, instantly.

Built around accessible 3D-printed hardware and off-the-shelf actuation, Bionikle targets a fraction of the cost of conventional powered prosthetics without giving up responsiveness or safety.

3D PrintedAccessible

Manufacturable hardware that keeps cost low and iteration fast.

On-device AIAdaptive

Real-time gait classification reacts to how you actually move.

Custom ControlResponsive

Finite-state control with continuous safety verification.

How Bionikle thinks and moves

From raw motion to actuated joint, every cycle runs on-device in real time.

FEEDBACK LOOPIMUAI ClassifierState MachineActuationSafety
Forward path Feedback loop

Data flow runs in sequence: IMU Sensing then AI Classifier then State Machine then Actuation then Safety Loop, with the safety encoder feeding back to the state machine.

  1. 01

    IMU Sensing

    Motion data streamed from an onboard inertial unit.

  2. 02

    AI Classifier

    An SVC model labels the locomotion mode: walk, stairs-up, or stairs-down.

  3. 03

    State Machine

    A finite-state controller selects the right joint trajectory for the moment.

  4. 04

    Actuation

    A high-torque TMotor AK80-9 drives the ankle joint over the CAN bus.

  5. 05

    Safety Loop

    A magnetic encoder verifies position continuously to keep motion safe.

real-time · on-device · closed-loop

Built on accessible hardware

Capable parts, chosen so the whole system stays affordable and repairable.

Compute

Raspberry Pi

Runs sensing, AI, and control on-device, with no cloud and no latency.

Actuator

TMotor AK80-9

High-torque quasi-direct-drive joint over a 1 Mbps CAN bus.

Encoder

AS5600 Magnetic

Absolute joint feedback for continuous safety checks.

Structure

3D-Printed Frame

Lightweight, low-cost, and fast to iterate on.

The team

The people building Bionikle.

AY

Angelo Yaghi

Mechatronics Engineer

LB

Lea Babi

Mechatronics Engineer

FK

Fadi Kaafarani

Mechatronics Engineer

DN

Dr. Noel Maalouf

Project Supervisor

Get in touch

Questions, collaboration, or just curious? Reach out.

Final-Year Engineering Project

Your University, Department of Engineering