This paper explores a set of locomotion optimization
methods for a novel tendon-driven robotic fish.
With the typical features of the dual tendon driving active
tail and variable stiffness passive caudal fin, a fully
functioning tendon-drive robotic fish is developed. Next, a
practical dynamic model for the tendon-driven robotic fish
is established with full consideration of variable stiffness
passive caudal fin, and the model parameters are accurately
identified via data-driven methods. More importantly, to
improve the motion performances, an asymmetric central
pattern generator (CPG) is particularly proposed, and an
adjustment rule of passive stiffness is explored to fit different
motion states of the robotic fish. Finally, extensive
simulations and aquatic experiments verify the feasibility
of the proposed prototype and locomotion optimization
methods. The obtained results show that the improvements
of steering radius and forward swimming velocity are 36.3%
and 29%, respectively. At present, our robotic fish can
achieve maximum forward swimming as 1.04 BL/s (BL for
the Body Length), maximum turning rate as 153.3◦/s, and
the minimum turning radius reaches to 0.31 BL, providing
a valuable reference for bio-inspired research of aquatic
mechanical design and locomotion control.