Highly Agile and Untethered Centimeter-Scale Swimming Robots for Autonomous Surface Navigation and Environmental Sensing

Significance 

The field of aquatic robotics has been pivotal in environmental monitoring, marine biology, and deep-sea exploration. Existing swimming robots have provided insights into pollution mapping, underwater biodiversity, and remote exploration. However, most robots in this domain suffer from critical limitations: large sizes (>30 cm), substantial weight (>1 kg), and inefficient propulsion strategies that confine them to open waters. Navigating complex environments, such as densely vegetated water surfaces, remains a significant challenge due to these physical constraints. Biomimetic approaches have attempted to address these limitations by designing robots inspired by aquatic species such as water striders and rays. While centimeter-scale robots (1–10 cm) have been developed to enhance maneuverability, they frequently require external power sources, are limited in mobility, or lack autonomous functionality. The reliance on tethered systems restricts real-world applications, particularly in cluttered environments where untethered navigation is essential. To overcome these challenges, new research paper published in Journal of Science Robotics and conducted by Dr. Florian Hartmann, Dr. Mrudhula Baskaran, Dr. Gaetan Raynaud, Dr. Mehdi Benbedda, Professor Karen Mulleners, and led by Professor Herbert Shea from the École Polytechnique Fédérale de Lausanne (EPFL) proposed a novel, highly agile, centimeter-scale flat swimming robot that operates untethered on the water surface. The proposed system features soft electrohydraulic actuators driving undulating pectoral fins, generating traveling waves for propulsion. Unlike conventional systems relying on external power, this design integrates a compact and lightweight power supply, reducing operating voltages to below 500 V and power consumption to 35 mW. The robot is designed to navigate confined spaces, autonomously detect and respond to environmental stimuli, and push objects 16 times its own weight, making it an ideal candidate for complex aquatic exploration and monitoring applications.

The researchers conducted extensive experimental evaluations to validate the performance of the proposed robotic system. The first set of experiments focused on locomotion efficiency and propulsion mechanisms. The robot’s pectoral fins, driven by a single electrohydraulic actuator per side, generated traveling waves on the water surface, creating propulsion. The experimental setup included both free-swimming tests and controlled force measurements using a glass cantilever. Results demonstrated that the robot achieved translation speeds of 5.1 cm/s in untethered operation and 12 cm/s when tethered. Additionally, the robot demonstrated high maneuverability with rotation speeds of up to 195°/s. A key innovation in this study was the integration of a miniaturized high-voltage power supply (HVPS), enabling untethered operation. The HVPS efficiently converted battery power to 500 V signals required for actuator control. The authors optimized actuator performance by selecting a ferrorelaxor terpolymer dielectric, reducing required voltages while maintaining high actuation bandwidths (>100 Hz). This material choice enhanced actuator longevity, with experimental results showing over 750,000 actuation cycles before performance degradation. They used scaling analysis experiments to study the effects of varying fin dimensions and actuation frequencies on thrust generation. The researchers derived an analytical model correlating optimal fin span with thrust efficiency, demonstrating that a 20-mm fin span maximized propulsion efficiency. The experiments revealed that larger fin spans increased thrust while requiring lower actuation frequencies. The results also showed that robots with smaller characteristic sizes required higher actuation frequencies to maintain optimal propulsion. The team conducted further experimental studies to assess the robot’s capability to navigate cluttered environments and perform obstacle interactions. The robot successfully swam through narrow spaces (~5 cm), circumvented obstacles, and pushed away floating objects weighing over 16 times its mass. Optical sensing capabilities were incorporated for autonomous navigation, with experiments demonstrating the robot’s ability to detect and respond to external light sources. When programmed for light-seeking behavior, the robot autonomously adjusted its trajectory to follow sequentially activated light sources. To extend the functionality, a four-actuator variant of the robot was tested. This configuration allowed for omnidirectional swimming, including forward, backward, and sideways motion, analogous to quadcopters in air. The four-actuator system maintained similar propulsion efficiency as the two-actuator design while introducing greater control flexibility. Performance characterization indicated a trade-off between power consumption and agility, with multi-actuator configurations requiring higher power inputs but enabling complex maneuvers.

In conclusion, the research work of Professor Herbert Shea and his colleagues is significant for aquatic robotics and environmental monitoring. The successful development of a highly agile, untethered centimeter-scale swimming robot paves the way for autonomous water surface exploration in complex environments. The combination of electrohydraulic actuation and optimized power management provides a scalable approach to designing efficient miniature robots for real-world applications. Potential applications include environmental monitoring, where these robots could autonomously measure water quality parameters in lakes, rivers, and aquaculture environments. The ability to navigate dense vegetation and interact with floating debris makes them ideal for tasks such as microplastic detection, pollutant mapping, and ecological monitoring. Additionally, the integration of optical sensing capabilities suggests broader applications in automated inspections and swarm-based robotic systems for large-scale water surface exploration. From a technological perspective, this research advances the field of soft robotics by demonstrating that electrohydraulic actuators can be effectively scaled down for efficient locomotion in water. The findings underscore the importance of material selection, actuator design, and power optimization in enhancing robotic mobility at small scales. Future developments could incorporate additional sensing modalities, improved energy efficiency through solar integration, and enhanced swarm coordination for cooperative exploration.

Highly Agile and Untethered Centimeter-Scale Swimming Robots for Autonomous Surface Navigation and Environmental Sensing - Advances in Engineering
schematic view of the untethered robot. Image credit: Sci Robot. 2025 Feb 19;10(99):eadr0721. doi: 10.1126/scirobotics.adr0721.

About the author

Associate Professor Karen Mulleners

École Polytechnique Fédérale de Lausanne (EPFL)

Switzerland.

Our lab focuses on the investigation of unsteady vortex dominated flow phenomena. We combine high spatially and temporally resolved flow field measurements with – what we refer to as – a fluid dynamical differential diagnosis. We develop methodologies to help bridge the gap between our observations and understanding of the development and interaction of vortices and technically relevant quantities, such as aerodynamic forces or power output, with applications in biological and bio-inspired flight, wind turbine rotor blade aerodynamics

About the author

Professor Herbert Shea

École Polytechnique Fédérale de Lausanne (EPFL)
Switzerland.

Research Interests:

We are widely recognized for our innovative fabrication techniques and elegant applications of elastomer-based transducers using electrostatic forces (eg DEA, HAXELS, EHD pumps) Our focus is on complex soft machines and soft robots.

We develop arrays of low-power flexible actuators for tactile and haptic displays for VR, AR and to allow blind and visually-impaired users to access dynamic graphical information.

We develop flexible sensors on cellulosic and polymeric substrates for a variety of applications, such as smart packaging, environmental and health monitoring, IoT, wearables and implants.  Physical, chemical and biological sensors have been demonstrated and integrated into single or multi-sensing systems incorporating microfluidic, electronic, and wireless functionalities, among others.

Reference

Hartmann F, Baskaran M, Raynaud G, Benbedda M, Mulleners K, Shea H. Highly agile flat swimming robot. Sci Robot. 2025 Feb 19;10(99):eadr0721. doi: 10.1126/scirobotics.adr0721.

Go to Sci Robot.

Check Also

Bridging Model Fidelity and Deep Learning: A Robust Reconstruction Framework for Lensless Imaging - Advances in Engineering

Bridging Model Fidelity and Deep Learning: A Robust Reconstruction Framework for Lensless Imaging