Graph Based Morphology Encoding for Robust Generalisation of RL Policies RL policies need to be re-trained and fine tuned everytime when being tested on a new design. This research focusses on encoding methods to assist in application of generalised policies over a new real world robot design in simulation Trajectory Optimisation for Non-Prehensile Manipulation Generated Jerk optimised trajectories for the xarm7 manipulator using Optimal Control for the Non-Prehensile manipulation task of transporting objects using a flat plate end-effctor.