When exoskeletons are driven in open loop with predetermined trajectories, the onus is placed on the user to maintain balance through their crutches. This work uses simulation of a human-exoskeleton model to explore the idea that such trajectories could be optimised to give the user the ‘best chance’ to maintain their balance in the presence of perturbations. The method evaluates a reward function under different gait trajectories and initial poses. It is concluded that such an optimisation method could increase the set of perturbations which a user can counter without adding significant complexity or expense to the exoskeleton.