As machine users generally only define the start and end point of the movement, a large trajectory optimization potential rises for single axis mechanisms performing repetitive tasks. However, a descriptive mathematical model of the mecha- nism needs to be defined in order to apply existing optimization techniques. This is usually done with complex methods like virtual work or Lagrange equations. In this paper, a generic technique is presented to optimize the design of point-to-point trajectories by extracting position dependent properties with CAD motion simulations. The optimization problem is solved by a genetic algorithm. Nevertheless, the potential savings will only be achieved if the machine is capable of accurately following the optimized trajectory. Therefore, a feedforward motion controller is derived from the generic model allowing to use the controller for various settings and position profiles. Moreover, the theoretical savings are compared with experimental data from a physical set-up. The results quantitatively show that the savings potential is effectively achieved thanks to advanced torque feedforward with a reduction of the maximum torque by 12.6% compared with a standard 1/3-profile
CAD enabled trajectory optimization and accurate motion control for repetitive tasks
2019-01-01
15th IEEE International Conference on Control and Automation
Conference paper
Electronic Resource
English
DDC: | 629 |
Fast and Accurate Integration of Launcher Trajectory Applied to Trajectory Optimization
British Library Conference Proceedings | 1998
|