Seahorse tail
Seahorses grasp seagrass and corals with a prehensile tail. Its hypaxial myomeres (HM) form long muscle-tendon
sheets that can span up to eleven vertebral segments, next to short median ventral muscles (MVM) between
neighbouring segments. The BRT seahorse is the in-silico tail of
Marzougui et al. (2025), a digital twin of a 3D-printed
prototype. It has no task: it is a playground for muscle architectures, driven with open-loop controls.
Sliders set how strongly each muscle group contracts (HM: fraction of the maximum force, MVM: fraction of the
beam's length it pulls in). Muscles light up red as they pull. Drag to orbit, right-drag or shift-drag to pan,
scroll to zoom. The meshes are a 3.4 MB download.
What you are seeing
The default seahorse/tail environment: 11 segments, each a vertebra inside four overlapping
plates, hanging from a fixed base with its ventral side to the right. Two force-controlled HM beams run
along the ventral side from the base to the tip, and every pair of neighbouring segments has two
position-controlled MVM beams. Contracting both HM beams curls the whole tail ventrally; contracting one
bends it sideways as well. The MVMs bend the tail locally. Contacts are off, as in the original model.
Gravity is switched on in this demo (it is off by default in Python) so that the tail hangs straight
again when its muscles relax.
Your own muscle architecture
In Python, the morphology specification sets the number of segments, which segments each HM beam spans
and which attachment points it passes, whether the MVM beams exist, and position or force control. The
reward is always 0 and episodes never terminate, so you drive the tail with your own open-loop controls.
Read the seahorse README, the
paper (J. R. Soc. Interface 22: 20240876), or
try the seahorse muscle playground
from the paper's repository.