Meet The Robotic Dog That Can Swim Just Like A Retriever

GUANGZHOU, China — Dogs are natural swimmers, instinctively knowing how to paddle when they hit the water. Now, engineers have created a robotic dog that can do the same, seamlessly transitioning from land to water without missing a beat.

A team of international researchers has developed a waterproof robotic dog that can trot on land and paddle in water, mimicking the swimming technique of real canines. The amphibious robotic dog (ARD) achieved a swimming speed of 0.16 meters per second (about 6 inches per second) and can move on land at more than twice that speed.

The research, published in the journal Bioinspiration & Biomimetics, builds on established designs and movement patterns from existing robotic dogs rather than starting from scratch.

How Dogs Inspired a Swimming Robot Design

Unlike previous amphibious robots that need special propellers, inflatable skins, or jet devices to move through water, this new robot uses the same four legs for both walking and swimming. It simply changes its gait pattern when transitioning between environments.

Engineers closely studied how dogs swim, analyzing the back-and-forth motion of their legs and how they extend their limbs during what’s called the “power phase” of paddling to generate thrust. They then programmed these movements into their robot.

Many mammals can swim at relatively high speeds when necessary for survival activities like hunting, migration, or escaping danger, particularly over shorter distances.

The team tested three different swimming patterns: two versions of what’s called a “lateral sequence paddling gait” (LSPG), where the legs move in a specific sequence like a dog’s natural swim, and a “trot-like paddling gait” (TLPG), where diagonal pairs of legs move together.

The Trade-Off Between Speed and Stability

The LSPG provided faster swimming speeds but less stability, while the TLPG offered greater stability in the water with slower speeds. This presents an important design consideration for future amphibious robots depending on their intended use.

The team found that creating effective amphibious robots requires finding the right balance between speed and stability, depending on the tasks the robot needs to perform. LSPG is preferable for speed-critical missions, while TLPG is better suited for operations requiring stability.

The robot’s design carefully considered the relationship between the center of gravity and center of buoyancy. Like a dog, the robot needed its lungs (or in this case, lighter components) positioned to keep it properly oriented in water. The team deliberately placed the battery and main control board near the bottom of the shell to lower the center of gravity and improve stability.

The researchers put their robot through rigorous testing with swimming tests in both controlled laboratory pools and natural lake environments. They even tested how well it could handle moving water by having it swim against currents of different strengths.

The amphibious robotic dog reached a maximum water speed of 0.16 meters per second (0.54 body lengths per second) and a land speed of 0.35 meters per second (1.16 body lengths per second). While not as fast as some specialized robots designed specifically for either land or water, it demonstrates impressive versatility across both environments.

The team also measured how much power the robot consumed in different modes and calculated the “cost of transport,” essentially how much energy is required to move a certain distance. Swimming with the LSPG consumed much less power than walking with a trot gait (23.26 watts versus 40.70 watts) and provided 46% better energy efficiency than the TLPG swimming mode.

Researchers were impressed with the robot’s ability to transition between environments. In one test, the team had the robot walk down a sloped surface into water, automatically switching from walking to swimming as it became submerged.

Future improvements could include adding more sensing capabilities and refining the control algorithms to handle more complex terrains. The researchers also mentioned the potential for exploring “underactuated multi-joint leg structures” that could further improve swimming efficiency.

Robots could one day seamlessly navigate across diverse environments without specialized equipment for each terrain type, just like many animals do. Advancements like these would help to make robots more useful for real-world applications like search and rescue operations.