Please enable JavaScript.
Coggle requires JavaScript to display documents.
Machine Learning for Prosthetics - Coggle Diagram
Machine Learning
for Prosthetics
User Intention Detection
Prosthesis Personalization
adaptation of
control parameters
to user preferences
to morphological properties
during first fit or
over the course of a day
Prosthesis Control
Expressing
Control Commands
by high-level 'commands'
think 'picking up'
and prosthetic
executes an appropriate grip
by Natural Language
Control Signals
target angles for
low-level PD control
joint torque
joint stiffness and damping
Environmental Features Recognition
lower-limbs
slope of road
width/height of stairs
terrain properties
upper-limbs
object detection
surface properties
Testing new Prostheses
with Deep RL in simulation
with musculoskeletal models
and Deep Reinforcement Learning
Data Types
ultrasound
forces
joint torques
spring forces
ground reaction forces
motion capture
metabolic cost
EMG
IMU
unconventional
data
computer vision
depth, distance
point cloud
limb deformation
movement between residuum and socket
kinematics of other joints
or the other limb
sound
strain gauges
temperature
pressure
State Detection
Falling Detection
User Fatigue