Energy efficient trajectories for an industrial ABB robot

Abstract This paper presents a systematic methodology for on-site identification and energy-optimal path planning of an industrial robot. The identification experiments are carried out on-site, in a quick, non-invasive way using a CA8335 Qualistar three-phase electrical networks analyser. Next, the collected data is compared with a parametrised dynamic robot model in an optimisation routine. This routine results in the specification of a parametric dynamic robot model. The model is used as a dynamic constraint for a model predictive control problem, where other physical constraints are added i.e. the limited workspace and the constraints on the joint velocities and accelerations. A sequential quadratic programming solver is used to minimise a mechanical energy based cost function. The resulting energy-optimal path is translated into custom robot commands executable on an industrial robot. The systematic methodology is validated on an IRB1600 industrial ABB robot performing a custom pick-and- place operation. The obtained dynamic robot model is given and compared to the collected measurements. To demonstrate the possibility of energy saving by ‘intelligently’ programming a robot trajectory the energy and time-optimal paths are generated taking all physical constraints into account. Simulation results show a significant time and energy improvement (up to 5%) compared to most trajectories generated by the ABB software. The most remarkable result is that the fastest energy-optimal trajectory turns out to be 4% more energy efficient and 3% faster then the commercially available fastest trajectory. Additional stand-still experiments show that activation of the brakes is favoured over an actuated stand-still from an energy point of view, assuming that the start-up time when releasing the brakes is limited.