Numerical robot kinematics based on stochastic and molecular simulation methods

Thomas Kastenmeier, Franz J. Vesely
1996 Robotica (Cambridge. Print)  
Multilink robot arms are geometrically similar to chain molecules. We investigate the performance of molecular simulation methods, combined with stochastic methods for optimization, when applied to problems of robotics. An e cient and exible algorithm for solving the inverse kinematic problem for redundant robots in the presence of obstacles and other constraints is suggested. This Constrained Kinematics Stochastic Optimization" CKSO method is tested on various standard problems. 1 5 be a
more » ... ems. 1 5 be a minimum. This leads to the Moore-Penrose generalized inverse matrix Since the solution qt of the inverse kinematic problem is not unique, additional constraints or requirements may b e i n troduced at will. These may involve circumventing obstacles, minimizing angular velocities or accelerations at the joints, or avoiding extreme stretched or folded con gurations of the multilink structure.
doi:10.1017/s0263574700019640 fatcat:td4xwvd4abbmbalp4hu5y54l7u