Articulated swimming creatures

Abstract

We present a general approach to creating realistic swimming behavior for a given articulated creature body. The two main components of our method are creature/fluid simulation and the optimization of the creature motion parameters. We simulate two-way coupling between the fluid and the articulated body by solving a linear system that matches acceleration at fluid/solid boundaries and that also enforces fluid incompressibility. The swimming motion of a given creature is described as a set of periodic functions, one for each joint degree of freedom. We optimize over the space of these functions in order to find a motion that causes the creature to swim straight and stay within a given energy budget. Our creatures can perform path following by first training appropriate turning maneuvers through offline optimization and then selecting between these motions to track the given path. We present results for a clownfish, an eel, a sea turtle, a manta ray and a frog, and in each case the resulting motion is a good match to the real-world animals. We also demonstrate a plausible swimming gait for a fictional creature that has no real-world counterpart.

A Comparison of Action Spaces for Learning Manipulation Tasks

“…In response to this, Tan et al [22] developed an implicit PD control scheme that can allow arbitrarily large gains. While this is a popular choice in the graphics community [23], [24], [25] and for learning in simulation [10], [26], [27], [28], [29], the control law is non-causal and does not extend to physical systems.…”

Interactive Character Animation Using Simulated Physics: A State‐of‐the‐Art Review

“…[YLvdP07]) implicitly embodies coordinated action. In parallel, research in off-line optimization of control parameters has resulted in impressive controllers, both for new behaviours [YCBvdP08], stylized locomotion [WFH09,WFH10], and swimming behaviours [TGTL11].…”

“…Most physics-based characters are modelled after humans, while some research papers model characters after animals [RH91, GT95, CKJ*11, TGTL11], robots [Hod91,RH91], or creatures that do not exist in nature [Sim94,TGTL11]. Character models are often simplified to increase simulation performance and to make them easier to control.…”

“…In recent research, Tan et al [TGTL11] use procedural motion generation and joint-space motion control to animate underwater creatures, using target trajectories based on geometric functions. The parameters for these functions are found through off-line optimization for high-level criteria, such as maximization of swimming speed and minimization of energy consumption.…”

Task-centric optimization of configurations for assistive robots

“…However, we found that uniform sampling had difficulty finding good configurations for tasks where the solution space was small. Tan et al (2011) used Covariance Matrix Adaptation Evolution Strategy (CMA-ES) to design a controller for articulated bodies moving in a hydrodynamic environment, which inspired our use of CMA-ES (from https://pypi.python.org/pypi/cma) for our optimization. We used a heuristic when both P R and P M are zero that pushes the search toward configurations that may have non-zero P R and P M .…”