Omnidirectional kick in RoboCup3D simulation

Abstract

Robocup competition requires robots with rapid reaction and efficient executive in the changing environment, which keeps team-attacking efficiently. This paper proposed a method based on CMA-ES optimization to accomplish robots' omnidirectional kick. The omnidirectional kick consists of path planning module, inverse kinematics module and optimization module. The path planning module designs the trajectory that the foot must follow to propel the ball in the intended direction. To ensure the robot's kick, the inverse kinematics module is responsible for computing the value of robots' leg joints. The optimization module takes the CAM-ES gradual accumulation learning optimization algorithm. The simulation experimental results show our method can make robots perform accurately predetermined behaviors and the ball kicked to a desired position.

A Novel Method for Inverse Kinematics Solutions of Space Modular Self-Reconfigurable Satellites with Self-Collision Avoidance

“…When adding additional constraints to the multi-solution inverse kinematics, the corresponding optimal solutions for different objectives can be obtained. Many researchers have attempted to solve inverse kinematics with a variety of meta-heuristic algorithms, including the Genetic algorithm (GA) [10,11], the Artificial Bee Colony algorithm (ABC) [12,13], the Particle Swarm Optimization algorithm (PSO) [14,15], the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) [16] and the Differential Evolution algorithm (DE) [17,18], et al…”