Mathematical optimization in robotics: Towards automated high-speed motion planning (Q1393003)

The paper aims at providing mathematical and mechanical models for adequate controlling of the complex nonlinear dynamics of very fast maneuvers of industrial robots. These models should overcome the drawbacks of traditional kinematic models which deal only with the kinematics of a robot and with rudimentary velocity and acceleration properties. The authors describe therefore the kinematic and dynamic modeling of the robot, deal with forward and inverse dynamics, and apply their model to an industrial robot, namely the KUKA 761. Furthermore they incorporate rotor inertia, joint elasticity, motor dynamics, friction and a pneumatic weight compensation into the dynamic model. A main objective is then the optimization of the model parameters. Two new sequential quadratic programming algorithms for trajectory optimization are presented and tested in an application example. This application is a time-critical transport maneuver. Although the detailed analysis shows only a gain of 9\% in the complete optimized maneuver, this gain has to be considered.