For linear and static cases, topology optimisation of structures is a well-established field. More recently, nonlinearities have been integrated into the approaches together with dynamic assessments. Here, topology optimisation of structural crashworthiness has been a challenging research question where new methods have been explored. The criteria for optimality in these achievements address mainly mechanical responses like compliance, stress, forces, weight, energy absorption, or acceleration. Considering the increasing need to design for sustainability, we lack here new ideas and methods. In most optimisations, weight reduction is the only established aspect related to ecological design. Therefore, first ideas will be presented here based on traceable machine learning techniques, which may allow integrating sustainability into optimisation techniques. They relate to multi-criteria optimisation, preference-driven designs, surrogate modelling, and machine learning. Criteria like similarity (similarity to a known design concept having higher sustainability), modularity related to reparability, re-usage, use-conversion, or recycling, and flexibility required for alternate usages/use-adaptations should become the driving aspects for optimisations.