High-Efficiency Closed-Loop Control of a Robotic Fish via Virtual Musculoskeletal Methodology

Abstract

Improving propulsion efficiency holds the promise of enabling the robotic fish to work for a long time with a limited battery in its small body. In this paper, for the swimming of a bionic robotic fish, we present a virtual musculoskeletal control method from the bionic model of the joint driven by agonist muscle and antagonist muscle. A closed-loop method composed of two loops is proposed as a rule of thumb for the speed control of the robotic fish. The outer loop adjusts the swimming speed using the speed deviation; the inner loop regulates the stiffness according to the virtual muscle spindle feedback to fit the water environment. Compared with the proportion control, the evaluation results show that the virtual musculoskeletal methodology increases the efficiency by 3.4% in the steady flow and 7% in the Karman-vortex flow. This algorithm provides a new idea for the joint-space control of the bionic robots that need to reduce the energy consumption of movements.