Abstract
Sensory feedback plays a very significant role in the generation of diverse and stable movements for animals. In this paper we describe our effort to develop a Central Pattern Generator (CPG)-based sensory feedback control for the creation of multimodal swimming for a multi-articulated robotic fish in the context of neurocomputing. The proposed control strategy is composed of two phases: the upper decision-making and the automatic adjustment. According to the upper control commands and the sensory inputs, different swimming gaits are determined by a finite state machine algorithm. At the same time, the sensory feedback is exploited to shape the CPG coupling forms and control parameters. In the automatic adjustment phase, the CPG model with sensory feedback will adapt the environment autonomously. Simulation and underwater tests are further conducted to verify the presented control scheme. It is found that the CPG-based sensory feedback control method can effectively improve the manoeuvrability and adaptability of the robotic fish in water.
Highlights
In recent years, more and more bio-inspired mechatronic systems that can operate in unstructured environments robustly and efficiently have been proposed
The obstacle-related sensory information was coupled into the Central Pattern Generator (CPG) control model to autonomously trigger a gait transition
Within our proposed two-phase CPG-based sensory feedback control model, if the robotic fish detected the wall of the pool, the gait of turning would be initiated
Summary
More and more bio-inspired mechatronic systems that can operate in unstructured environments robustly and efficiently have been proposed. A precondition behind this trend is that long evolved biological solutions can partially or fully be transferred to engineering systems [1]. In the context of neurocompting, the CPGs can be considered as a dedicated neural mechanism involving a group of neurons that generate rhythmic signals in a coordinated manner without sensory feedback, while sensory feedback is needed to shape the CPG signals. This kind of neural mechanism generally underlies the production of most rhythmic motor patterns, such as swimming, walking, and hopping. CPGs are extended to cover discrete motions as well [7]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
