Porous 3D-Printed Robot Feet Enhance Energy Efficiency
Researchers have developed 3D-printed porous feet for quadruped robots that, when combined with a learning-based controller, reduce energy consumption by up to 6.2%.
Quadruped robots are increasingly deployed for tasks like inspection and search-and-rescue, but their leg-based locomotion is significantly less energy-efficient than wheeled alternatives. A common strategy to improve efficiency involves incorporating elastic components in the legs to store and release impact energy. However, at lower speeds, this stored energy can be lost as heat, reducing effectiveness and potentially destabilizing the robot.
To address this, researchers at Seoul National University of Science and Technology have designed and 3D-printed porous triply periodic minimal surface (TPMS) robot feet. These feet are engineered to store and release impact energy. The team paired these compliant feet with a deep reinforcement learning controller that was trained to optimize the exploitation of the stored elastic energy during walking.
The study explored three TPMS hemispherical foot designs: primitive, gyroid, and diamond. Through compression tests, the diamond design with a 60% relative density was selected for its superior flexibility, energy absorption capabilities, and minimal energy dissipation. This design choice was crucial for maximizing the benefits of passive energy management.
To enable the robot to effectively utilize the energy-storing properties of the TPMS feet, a deep reinforcement learning controller was employed. This controller analyzed various walking strategies, considering the energy consumption and the specific way the TPMS feet compressed and rebounded. The system then coordinated the robot's gait with the feet's passive energy dynamics, thereby reducing the workload on the motors. When tested on a commercial robot, this integrated system demonstrated a reduction in battery power consumption ranging from 1.4% to 6.2% at walking speeds between 0.4 and 1.0 meters per second.
This development showcases the integration of advanced additive manufacturing (TPMS structures) with intelligent control (deep reinforcement learning) to improve robotic energy efficiency. By creating compliant, energy-recovering feet, it offers a novel passive mechanism to reduce motor workload, a critical factor for extending operational range and reducing the need for frequent recharging in mobile robotics.
Edited by the news editor with AI from the original report — please refer to the original source.