4 Conclusions
In our exploration of the problem of shape control, we have introduced a practical taxonomy to enhance the characterisation and comparison of shape control methods. Through criteria covering system characteristics, available information, error definition and control strategies, we have established a clear and systematic framework that not only facilitates the examination of existing methods, but it is also designed to embrace future innovations.
This taxonomy enabled us to identify several promising research directions. Firstly, there is a need to move beyond the commonly applied quasi-static manipulation assumption, as demonstrated in recent work (Fonkoua et al., 2024), particularly for tackling the unexplored problem of shape trajectory control. Another area for future research involves studying the use and positioning of soft grippers for shape control, exploiting continuous contact regions rather than the traditional single contact points (Cuiral-Zueco et al., 2022). Additionally, research may focus on the definition and feasibility analysis of target shapes, as current methods often define achievable targets by pre-acquiring shapes in experimental setups. Finally, reducing dependence on offline exploration of object behaviour (e.g., by exploiting more versatile models (Arriola-Rios et al., 2020)) is essential for advancing practical applications.