Figure 12.

This figure shows the influence of model learning on the 4-link balancing task. Illustrated are mean values and standard deviations over 20 runs. The learning performance with the given system model is denoted by PMP. Instead of using the given model we now want to learn the system model from data (as described in section 3.3), which is denoted by PMP-M. In the beginning of learning the extrinsic costs are larger compared to motor skill learning with a given analytic model. However, as the number of collected data-points $\langle [x_t;u_t],{\dot{x}}_t\rangle$ increases the PMPs with model learning quickly catch up and converge finally to the same costs. The PMP representation with model learning in parallel considerably outperforms the trajectory-based DMP approach in learning speed and in the final costs.