💻 computer science

A Sensorised Lattice Footplate for a Semi-Active Prosthetic Foot

This paper presents and validates a low-cost, semi-active prosthetic foot prototype that successfully embeds magnetic plantar sensors directly into a 3D-printed load-bearing lattice footplate to enable real-time force estimation and adaptive hydraulic damping control.

Original authors: Jinze Ge, Jingcheng Sun, Chengxu Zhou

Published 2026-06-25

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jinze Ge, Jingcheng Sun, Chengxu Zhou

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a prosthetic foot that doesn't just sit there like a rigid block of wood, but actually "feels" the ground and adjusts its own stiffness in real-time, all without needing a heavy battery or a complex motor to push you forward. That is the core idea behind this research paper.

Here is a breakdown of what the researchers built and discovered, using simple analogies:

The Problem: The "One-Size-Fits-All" Foot

Most current prosthetic feet are like stiff springs. Once they are made, their stiffness is fixed. If you walk on a flat sidewalk, they work okay. But if you walk up a hill, run, or step on uneven ground, they can't adapt. They can't change their "mood" to match the terrain.

Active robotic feet exist (like a car with an engine), but they are heavy, expensive, and complicated. The researchers wanted something in the middle: a semi-active foot. Think of it like a bicycle with a suspension system that you can manually tweak while riding, rather than a car with an engine. It doesn't generate its own power, but it can change how it absorbs shock.

The Solution: A "Smart Sponge" Foot

The team built a prototype with three main parts:

The "Smart Sponge" (The Lattice Footplate):
Instead of a solid piece of carbon fiber, they 3D-printed the bottom of the foot like a honeycomb or a sponge. This structure is flexible and can be squished.
- The Magic Trick: They embedded tiny magnets and sensors inside this sponge. As the foot squishes down when you step, the magnets move slightly relative to the sensors. The computer reads these tiny magnetic shifts and calculates exactly how much force is being applied.
- Why it matters: Usually, you need a separate insole (like a pressure mat) to measure this. This design puts the "feeling" mechanism inside the structure itself, making the foot compact and self-contained.
The "Adjustable Shock Absorber" (The Damper):
The foot has a hydraulic damper (like the shock absorber in a car) connected to an ankle joint.
- A small motor (servo) turns a gear that changes the "tightness" of this shock absorber.
- If the foot senses you are landing hard, it can tighten the damper to stop the foot from collapsing too fast. If you are rolling over your foot, it can loosen up to let you move smoothly.
The "Brain" (The Control System):
A computer reads the magnetic sensors, figures out where the weight is (heel vs. toe), and tells the motor how to adjust the damper.

What They Tested (The "Lab" Results)

The researchers didn't test this on a person walking yet. Instead, they put the foot in a machine and tested it in three ways:

Test 1: Does the sponge work?
They squished the 3D-printed lattice with a giant machine. They found they could change how stiff the sponge was just by changing the size of the holes in the honeycomb pattern. Crucially, the sensors inside the sponge accurately measured the force, even though they were buried inside the structure. It was like having a thermometer inside a cake that could tell you exactly how much pressure the oven was putting on the cake.
Test 2: Can it tell the difference between stances?
They manually held the foot in four different positions:
1. Heel Strike: Only the heel is touching.
2. Flat Foot: The whole foot is touching.
3. Dorsiflexion: The foot is bent up (toes up).
4. Toe-Off: Only the toes are touching.
  The sensors successfully told the difference between these four states. It's like the foot could say, "I know I'm on my heel right now," or "I know I'm on my toes."
Test 3: Does it move like a real ankle?
They simulated a walking cycle. The foot successfully mimicked the "rolling over" motion of a human ankle (dorsiflexion) by adjusting the damper.
- The Catch: The paper admits a major limitation. The damper can only resist motion (like a brake); it cannot push (like an engine). So, while the foot rolled over smoothly, it couldn't provide the final "kick" or "push-off" that a real human foot does to propel you forward. It's a great shock absorber, but it's not an engine.

The Bottom Line

This paper proves that you can build a low-cost prosthetic foot where the sensors are hidden inside the load-bearing structure (the 3D-printed lattice) rather than being a separate layer on top.

Cost: The prototype cost about $200 to make, compared to thousands for commercial versions.
Weight: It weighs about 0.8 kg, which is close to the weight of a human foot.
Verdict: It works as a research platform to test how a foot can "feel" and "adjust" its damping. However, it is not yet ready for a human to walk on, because it hasn't been tested on a treadmill, and it still lacks the ability to generate active power for pushing off.

Think of this as a proof-of-concept for a "smart" foot that can feel the ground and adjust its own shock absorption, paving the way for cheaper, more adaptable prosthetics in the future.

Technical Summary: A Sensorised Lattice Footplate for a Semi-Active Prosthetic Foot

Problem Statement
Current lower-limb prosthetic feet face a trade-off between adaptability and practicality. Passive energy-storing-and-returning (ESR) feet are robust and affordable but lack the ability to modulate stiffness and damping according to gait phase, speed, or terrain, potentially leading to compensatory gait strategies and increased loading on the intact limb. Conversely, active powered ankles offer superior biological mimicry but introduce significant mass, cost, and complexity due to motors, batteries, and control systems. Furthermore, existing adaptive solutions often rely on specialized mechanisms or external sensing layers (e.g., insoles) that are separate from the load-bearing structure, limiting compactness and integration. There is a need for a low-cost, semi-active solution that integrates sensing directly into the compliant element to enable damping regulation without the burden of active power generation.

Methodology
The authors propose and characterize an integrated prototype that combines three core components:

Sensorised 3D-Printed Lattice Footplate: Instead of a monolithic carbon-fibre spring, the footplate is a Body-Centred Cubic (BCC) lattice printed from PEBA. It embeds permanent magnets within predefined pockets and a magnetometer PCB beneath the lattice. This design utilizes magnetic tactile sensing (inspired by eFlesh) to estimate plantar contact forces and loading distributions based on deformation-induced magnetic field changes. A Multilayer Perceptron (MLP) processes 15-dimensional magnetic input vectors to regress force estimates.
Servo-Adjustable Hydraulic Damper: An articulated ankle mechanism couples a hydraulic damper to a servo-driven gear transmission. The servo adjusts the damper's valve opening, modulating the damping coefficient ( $b$ ) to regulate ankle resistance during the stance phase. The relationship between servo angle and damping coefficient was experimentally characterized and fitted to an exponential model.
Reduced-Order Dynamic Model: A sagittal-plane linkage model was developed to simulate the ankle's rotational dynamics, incorporating the variable damper and a return spring. This model was used to evaluate the system's ability to track biological ankle trajectories under a feedforward damping schedule.

The validation process involved three stages: (1) mechanical and sensing characterization of the lattice under controlled compression; (2) static-posture trials under body-weight loading to verify the end-to-end sensing pipeline's ability to distinguish stance configurations; and (3) simulation of the stance-phase response against a literature reference trajectory.

Key Contributions

Integrated Sensing Architecture: The paper demonstrates the embedding of magnetic plantar sensing directly into the load-bearing compliant element of a prosthetic foot, eliminating the need for a separate external insole sensing layer.
Reduced-Order Modeling and Characterization: It provides a coupled framework of a reduced-order articulated-ankle model and experimentally characterized damper dynamics, enabling simulation of the proposed semi-active mechanism.
Prototype Validation: The work presents a functional prototype that successfully tunes lattice stiffness via unit-cell geometry, estimates force with high fidelity under cyclic loading, and separates forefoot/rearfoot loading signatures across four static stance configurations.

Results

Lattice and Sensing Performance: Mechanical testing confirmed that footplate stiffness is tunable via unit-cell size. Under cyclic compression, the embedded sensor tracked the reference force from a universal testing machine with a Root Mean Square Error (RMSE) of 32.14 N (4.59% full scale), validating that force estimation is feasible within the load-bearing structure.
Static Posture Classification: In trials simulating heel-strike, foot-flat, dorsiflexion, and toe-off under approximately body-weight loading, the system produced linearly separable forefoot and rearfoot force distributions. A rule-based classifier correctly identified all four imposed postures, confirming the sensing pipeline's ability to distinguish loading patterns.
Simulation of Stance Response: The reduced-order simulation, driven by a feedforward damping schedule, successfully reproduced the dorsiflexion trend of a reference biological ankle trajectory during early-to-mid stance. However, as expected for a purely dissipative system, the simulation could not generate the active push-off observed in biological ankles during late stance, causing the trajectory to diverge from the reference.

Significance and Claims
The paper claims to demonstrate that magnetic plantar sensing can be successfully embedded inside the load-bearing compliant element of a low-cost prosthetic foot and that the resulting force readout is sufficiently informative to support a preliminary semi-active damping strategy. The authors position this work as a proof-of-concept for a compact research platform that integrates load-bearing support, adjustable damping, and plantar sensing.

The authors maintain a modest stance regarding the current limitations:

The sensing performance was evaluated only under controlled vertical loading; effects of shear forces, uneven contact, and material hysteresis remain unquantified.
The static-posture classifier is not validated for dynamic gait phase transitions.
The inability to generate active push-off is identified as a fundamental limitation of the semi-active (dissipative) architecture rather than a flaw in the prototype.

The study concludes that while the prototype is not yet a clinically deployable product (due to lack of certification and durability testing), it establishes a viable sensing-actuation pairing for future research into gait-dependent ankle damping modulation. The estimated cost of the prototype ($200) is significantly lower than commercial alternatives, though the authors note this comparison is between an uncertified research device and certified clinical products.

The Problem: The "One-Size-Fits-All" Foot

The Solution: A "Smart Sponge" Foot

What They Tested (The "Lab" Results)

The Bottom Line

More like this