Go with the Flow

Rotors, propellers, wheels, treads, and legs all have a way of quickly draining the precious energy reserves stored onboard a robotic system. For robots with long-distance missions — like those used for exploration, mapping, or search and rescue operations — that is a big problem. Once the juice is gone, it is lights out. For this reason, innovations in energy efficiency are urgently needed. Without technological advancements, autonomous systems that could otherwise reshape the world around us for the better will remain impractical for many real-world applications.

Researchers at Caltech have taken a meaningful step toward solving this challenge with a new underwater robotic system that “surfs” on naturally occurring fluid flows to propel itself forward with minimal energy expenditure. This novel method of propulsion could extend the operational range and longevity of autonomous underwater and aerial vehicles, bringing about new possibilities for exploration in challenging environments.

The palm-sized robot, known as CARL (Caltech Autonomous Reinforcement Learning robot), demonstrates how intelligent navigation can dramatically reduce the energy demands of autonomous systems. In particular, CARL exploits vortex rings — circular currents of water — to “ride” background flows rather than fight against them.

The team tested CARL in a 13,000-liter water tank equipped with a wall-mounted thruster to generate vortex rings. The robot was fitted with an onboard inertial measurement unit and ten motors that allowed it to move freely in all directions. However, instead of continuously propelling itself forward, CARL was programmed with a simple but highly effective algorithm that runs on a Teensy 4.1 development board: when it sensed a sudden acceleration in a specific direction — caused by the movement of the vortex — it would execute a short burst maneuver to enter the flow. Once inside, it could coast across the tank with minimal additional effort.

By taking advantage of these naturally occurring fluid structures, CARL was able to traverse the tank using only one-fifth of the energy that a conventionally programmed robot would require. This significant reduction in power consumption shows how autonomous systems can leverage environmental forces rather than work against them, potentially extending their operational capabilities by a large margin.

Today, autonomous underwater vehicles (AUVs) play a crucial role in ocean exploration, ecological monitoring, and underwater mapping. By harnessing currents in the ocean, future AUVs could extend their range and operate for longer durations without requiring frequent recharging or refueling. Similarly, aerial drones navigating complex wind patterns in urban environments could use gusts of wind to their advantage rather than continuously battling against them, conserving energy for mission-critical tasks.

Nature has long demonstrated the benefits of energy-efficient movement. Fish, for example, instinctively exploit unsteady vortices to swim with minimal effort, and birds use updrafts and wind currents to glide for extended periods. CARL’s ability to mimic these strategies in a robotic system suggests that future autonomous vehicles could become far more efficient by simply listening to what nature has been telling us all along.