Engineers Can Use This New Theory To Build Sprinting Robots

By studying how legged animals move across loose ground, like sand or gravel, researchers from the Georgia Institute of Technology were able to design a small robot that can run across these squishy, grainy surfaces with optimal efficiency.

From a practical standpoint this development is important for roboticists and engineers who want to design vehicles, like search-and-rescue robots or future Mars rovers, that can operate at their best speed on off-road environments.

After all, robots are the future.

Terradynamics

Engineers have historically used theories of aero- and hyrdo-dynamics to understand and predict how animals move on solid ground, by measuring lift, drag, and thrust forces.

The models not only describe how birds fly through the air or fish swim in the water, it has also helped designers build better airplanes or swimming robots. Basically any vehicles that are meant to go in fluids, like air and water.

Theories that predict how legged animals and robots move across a granular environment — surfaces made up of lots of small particles like sand and gravel — have proven more difficult than motion through air, water, and flat surfaces.

Environments like sand, rubble, snow, grass, or leaves, are more complex than solid ground because legs will sink.

Legged animals and robots that move across a granular surface have demonstrated that with each step, each part of the leg "moves through the substrate at a specific depth, orientation and movement of direction, all of which can change over time," according to a study published online Thursday, March 21, in the journal Science.

The area of terradynamics was developed so that researchers could predict the mobility of devices over these mushy or pebbly materials.

Developing a better robot

One feature that allows animals to move so well is that they have legs that can perform many functions.

"Legs allow the animal to climb over ledges, sprint over hard ground, paddle through soft ground and potentially kick through fluids," said Daniel Goldman, assistant physics professor at the Georgia Institute of Technology.

For this study, researchers started out with a simple toy robot. They took of the legs and replaced them with legs of different shapes that were 3-D printed.

Each leg is made up of little plate elements. A robotic arm uses a sensor to measure the force of all those little elements, which adds up to give scientists the total force of each leg. All of this is done before the legs are placed on the robot.

The researchers can predict the performance of a model by plugging this data, from different-shaped legs, into a special simulation called a multi-body simulation. The model can predict how robots will move over grainy surfaces.

They learned that "C"-shaped legs worked better than those that are saddle or Pringle-shaped, for example.

In the future, this information can help engineers optimize limb shape to create better small robots to explore unknown environments, like Mars, or search-and-rescue missions, where the ground may not be solid.

In time, we might be able to apply this framework for how little devices move across sand to bigger vehicles, Goldman said.