Trajectory optimization with GA and control for quadruped robots

Abstract

This paper proposes an optimal galloping trajectory, which costs low energy and guarantees the stability of the quadruped robot. In the realization of fast galloping, the trajectory design is important. For a galloping trajectory, we propose an elliptic leg trajectory, which provides simplified locomotion to complex galloping motions of animals. However, the elliptic trajectory, as an imitation of animal galloping motion, does not guarantee stability and minimal energy consumption. We propose optimization based on energy and stability using a genetic algorithm, which provides a robust and globally optimized solution to this multi-body, highly nonlinear dynamic system. To evaluate and verify the effectiveness of the proposed trajectory, a series of computer simulations were carried out.

Horse-like walking, trotting, and galloping derived from kinematic Motion Primitives (kMPs) and their application to walk/trot transitions in a compliant quadruped robot

“…Quadrupedal locomotion presents good intrinsic stability features, but the coordination of the motion of four legs is non-trivial to control. Different trajectory generation techniques have been adopted (Kimura et al, 1989;Raibert, 1990;Sakakibara et al, 1990;Kramy and Orin, 2003;Golubovic and Hu, 2003a,b;Tsujita et al, 2005;Hu and Gu, 2005;Iida et al, 2005;Kim et al, 2006;Rebula et al, 2007;Pongas et al, 2007;Hebbel et al, 2007;Chae and Park, 2008;Kim et al, 2008), and the efforts in this field of research led to proficient results in the development of different gaits, from the slow walk to gallop. Raibert (1986) provides a good overview of legged locomotion, with particular attention given to quadrupeds.…”

Prediction of building energy consumption using an improved real coded genetic algorithm based least squares support vector machine approach

“…Mutation is an operator that introduces variations into the individual. This operation is performed occasionally with a regular probability [42]. Through the mutation operator, the search space is explored globally by searching for better points.…”

“…The process continues until the population converges to the global maximum or another stopping criterion is met. The procedure performed by the conventional RCGA is detailed below [42][43][44][45][46][47].…”

Optimization of bending sequence in roll forming using neural network and genetic algorithm

“…The new population is then used in the next iteration of the algorithm. The algorithm terminates when either a maximal number of generations has been produced or a satisfactory fitness level have been reached for the population [17].…”