Dendritic Gated Networks for Robust Prosthetic Control under Data Distribution Shifts

Name: Dendritic Gated Networks for Robust Prosthetic Control under Data Distribution Shifts
Uploaded: 2026-07-21T18:33:13.783Z
Duration: 58 min 44 s
Channel: Amii
Description: Research on dendritic gated networks for robust prosthetic control that better handles muscle signal variations than clinical standards. The approach combines neural networks with anatomical simulations and real-time validation in virtual reality.

Amii21 July 2026Watch on YouTube

Description

High abandonment rates for upper-limb prostheses are often driven by a loss of user trust when control systems fail to handle the variations of daily life. While current clinical standards like linear discriminant analysis (LDA) provide steady-state reliability, they struggle to adapt to real-world signal shifts caused by muscle fatigue or electrode movement. This seminar presents the use of dendritic gated networks (DGNs), a neural network architecture designed for robust, context-aware intent prediction directly on wearable edge devices. To address clinical data scarcity, we developed an anatomically-informed protocol to simulate muscle activation patterns, allowing for faster testing and validation of new control models. This talk details offline benchmarking showing that DGNs outperform LDA in dynamic conditions, particularly during electrode shifts and different limb positions, while improving safety by correctly identifying wrong predictions. Furthermore, we show results from real-time, human-in-the-loop virtual reality experiments where DGN-based control achieved higher task success rates and greater subjective reliability than the clinical standard. Finally, we outline future work exploring DGNs for continual learning on hardware with strict resource constraints and their extension to other robotic environments. BIO: Laura is currently pursuing a Ph.D. in Computing Science under the supervision of Dr. Patrick Pilarski (BLINC lab) and Dr. Matthew E. Taylor (IRL lab). She received a B.Sc. with Honors in Computing Science from the University of Alberta in 2019 and an M.Sc. in Computing Science from the University of Alberta in 2022. Her research interests include neuroprosthetics, machine learning, and human-robot interaction. Drawing inspiration from her anatomical studies, Laura’s research aims to develop robot control methods with the goal of increased functionality, reliability, and safety in the real-world.

What you'll learn

Dendritic gated networks (DGNs) are a neural network architecture that makes prosthetic control more robust against variations in muscle signals and electrode movements.
Anatomically-informed simulations of muscle activation patterns accelerate testing of control models without requiring extensive clinical data.
DGNs outperform the current clinical standard (LDA) in dynamic conditions and safely detect when predictions fail.
Virtual reality experiments with humans show DGN-based control achieves higher task success rates and greater user confidence than standard methods.
The research focuses on continual learning on edge devices with limited compute power for prostheses in everyday life.

Frequently asked questions

Why do current prosthetic control systems fail with variations in muscle signals?

Clinical standards like linear discriminant analysis (LDA) provide steady-state reliability under fixed conditions, but cannot adapt to real muscle signal changes caused by fatigue, electrode movement, or different body positions.

How do anatomically-informed simulations help develop better prosthetic control?

Anatomical simulations allow researchers to replicate muscle activation patterns without extensive clinical data. This enables faster and more cost-effective testing of new control models before real-world validation.

What advantages did dendritic gated networks show in virtual reality experiments?

In VR tests, DGN-based control achieved higher task success rates and greater user confidence compared to the clinical standard, indicating users trust the system more.

What is the next step in research on DGNs for prosthetics?

Future work focuses on continual learning on hardware with strict resource constraints and extending DGNs to other robotic environments beyond prosthetics.