Energy-based Control of Soft Robots

Soft limb locomotion is a relatively new and challenging research field. These robots are a good example of the much wider class of bio-inspired soft robots, with the particular advantage of being easy and inexpensive to build and field. Soft limbed limbs belong to the class of underactuated systems and exhibit higher inherent compliance and stiffness modulation than rigid limbed robots and can tolerate a wider range of impacts without compromising either safety or stability. Moreover, minimal ground impact forces are expected because of their light weight construction and inherent compliance.Contrary to rigid robots, soft limbed robots are subjected to elastic deformations that increase the control complexity. Physically, these soft robotic limbs enable new and robust modes of terrain adaptation with relatively simple and inexpensive structures. However, soft limbs can not yet transition to practical application due to difficulties associated with control methods. Traditional control strategies are faced with the daunting task of efficiently computing the control inputs under model uncertainty due to the elastic nature of the construction materials. Soft limb dynamics have to be derived through energy based methods to account for prominent potential energy interaction due to elastic materials used in construction. Hence, it makes practical senses to employ energy based control strategies for soft limbed robots. Energy based control approaches also have the advantage of shaping the system energy so that the resulting closed-loop system has the same physical structure as the original system hence consuming less control effort and being more safe and efficient.

The second figure shows the schematic of the underactuated compass gait soft limbed biped. Each limb has a linear elastic spring at the distal end to provide soft passive compliance. Equations of motion are derived through Euler-Lagrangian formulation and Port-Controlled Hamiltonian (PCH) formulations and then energy based controllers are designed. The figure below compares the stable limit cycles of passive compass biped robots of identical mechanical limb lengths and masses, but having rigid and soft compliant limbs, respectively. It is seen that the soft biped gives a larger range of limit cycle, and hence larger step lengths and velocities with some smooth dents in the shape due to the spring interaction.


Research Members: Isuru Godage, Yue Wang, and Ian D. Walker.

The following video compares the rigid and soft, underactuated biped walking side by side.

The following videos compares the lowest (top track) and fastest (bottom track) achievable speeds through kinetic energy shaping against the passive walking speed (mid track).


  1. Isuru S. Godage, Yue Wang, and Ian. D. Walker, “Energy Based Control of Compass Gait Soft Limbed Bipeds”, in IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2014, pp. 4057-4064.