- Published on
SeoulTech Researchers Develop AI-Designed Footpads that Cut Quadruped Robot Energy Use
- Authors
- Name
- DeepHorus
Researchers at Seoul National University of Science and Technology (SeoulTech) have developed 3D-printed footpads that reduce a quadruped robot’s battery consumption by as much as 6.2%, offering a cost-effective way to extend operating time without adding springs or other bulky hardware to the legs. The team combined flexible, porous footpads with an AI controller that learns to harness energy absorbed when the foot hits the ground. By timing the gait carefully, the system releases some of that stored energy during push-off, lessening the load on the robot’s motors.
The study, published online May 5 in the International Journal of Precision Engineering and Manufacturing-Green Technology, was led by professors Jung-Yup Kim and Keun Park. Four-legged robots typically consume far more energy than wheeled machines because they must repeatedly lift and place their legs. Previous attempts to recover energy using springs or elastic components have struggled, as these parts often lose stored energy at lower speeds and complicate control.
The SeoulTech team instead focused on the feet. They designed triply periodic minimal surface (TPMS) structures—lightweight, open-pore lattices that flex under pressure, absorb impact, and spring back. After 3D-printing three footpad variants (primitive, gyroid, and diamond patterns), they selected a diamond-pattern design with 60% relative density, which offered the best balance of flexibility, absorption, and low energy loss.
But a softer foot was not enough. The robot also needed to adjust its movement to match the foot’s compression and rebound. The researchers trained a controller using deep reinforcement learning, allowing the AI to discover gaits that leverage passive energy return. In tests at walking speeds of 0.4 to 1 meter per second, the optimized footpads cut battery use by 1.4% to 6.2% while maintaining stable locomotion.
The efficiency gains could benefit quadruped robots used in inspection, logistics, indoor services, and search-and-rescue, where longer battery life translates into greater range and less frequent charging. Professor Park noted, “By modulating foot stiffness and leveraging passive energy absorption and release, the proposed approach offers a practical alternative to conventional leg-mounted spring mechanisms.”
The work demonstrates how AI-driven control and advanced manufacturing can work in tandem to improve robotics efficiency. Going forward, the team plans to explore additional lattice geometries and adaptive controllers that adjust to varied terrains, potentially making legged robots more viable for real-world deployments. As the demand for autonomous mobile robots grows, this low-complexity solution could become a standard design element in next-generation quadruped platforms.
