Mechanically adaptive robotics: NSF CAREER Award
More capability without more power
When a robot needs to do more, the usual answer is a larger motor and a larger battery. The bicycle points to a different answer. A bicycle has no motor, yet a person on a bicycle travels much farther and faster than on foot: the machine adds no energy, it changes how the human body uses energy. Mechanically adaptive robotics follows this idea. Instead of adding power, robots and wearable devices adapt the mechanics to the task, so that the same energy achieves more. The goal of this National Science Foundation CAREER Award is a theoretical foundation for such machines: robots and exoskeletons that provide capabilities beyond those of an unassisted human, to an extent previously thought to require large powered actuators.
Programmable springs
The key element is a new kind of mechanical component: the programmable spring. A programmable spring combines the benefits of a lever and a spring. The stiffness can be changed at near-zero energy cost, no matter how much energy the spring already holds, so the energy stored in the spring is no longer tied to how far the spring is compressed.
This opens up new ways to use energy. On a trampoline, a person builds up energy for a high leap through a series of ever-higher jumps. A robot limb with a programmable spring could store the same energy through a series of small, fixed-depth squats, a safer path to the same leap.
What the research has shown
The project built and tested devices that collect energy gradually and release it at the right moment:
- An actuator with an adjustable equilibrium position: a large direct-drive motor and a 3D-printed torsional spring, combined with a small motor that continuously shifts the equilibrium position, provide high torque-control bandwidth, energy storage, and shock mitigation in a single actuator (IEEE RA-L, 2026).
- A human-driven robot leg: with a pedaling mechanism, a person stores energy in the leg and increases the stiffness before the energy is released, amplifying both force and power, much like shifting gears while pedaling a bicycle (IEEE RA-L, 2026).
- Magnetic-spring actuators that reduce energy use in repetitive tasks by up to 50% compared with a conventional motor (IROS 2024).
- A floating spring leg that performs a demanding lifting task with no net energy input, a task not possible with ordinary fixed-stiffness springs (IEEE RA-L, 2023).
- Actuators with a variable stiffness spring in parallel with the motor, combining precise force control with efficient energy storage (IEEE T-RO, 2023).
- Energy accumulators that lock stored energy at any spring deflection and release it in less than ten milliseconds (ICRA 2023).
- A spring that fills up step by step: the weight of a person or robot compresses the spring a little more with every squat, under strict limits on force and motion (ICRA 2023).
- A spring with user-selectable stiffness, adjusted much like the gears of a bicycle (ICRA 2023).
- A theory of fast walking showing that a human-driven exoskeleton could bring a person carrying an extra body weight to race-walking speed within ten steps (IEEE TNSRE, 2022).
Where it could lead
- Assistive devices that help older adults move independently and safely.
- Exoskeletons that augment the physical performance of emergency responders.
- Unpowered legged devices that offer an alternative to bicycles for efficient transportation.
- Mobile robots that use less energy and operate longer between recharges.
This work was supported by the National Science Foundation CAREER Award “Mechanically Adaptive, Energetically Passive Robotics” (Award Number 2144551) under the Foundational Research in Robotics program within the Division of Civil, Mechanical, and Manufacturing Innovation.