Robots with instincts

This innovation can change everything in field robotics

Imagine a robot that does not depend on vision to find its way, feels the ground, listens to the vibrations of the environment and decides the best route as if it had instincts. Did you imagine the Boston Dynamics Atlas? But no, I'm not talking about it yet but rather about what researchers at Duke University are developing, a technology called Wild Fusion that allows robots to navigate rugged terrain as if they were humans.

The WildFusion structure was created to break a historical limitation of robots, vision as the only reliable sense and instead of relying only on cameras and Lidar sensors, this system combines multiple sensory sources to build a complete panorama of the environment. The quadruped robots that carry this technology have an RGB camera, Lidar, inertial sensors, contact microphones and touch sensors, all working in perfect harmony.

The auditory sensors capture the vibrations of the ground while the robot walks, the tactile sensors for their part measure the force applied to each leg allowing the robot to feel the ground under its feet, with this information it understands whether it is stepping on soft grass, on loose stone or on slippery gravel, all this sensory arsenal is encoded by a deep learning system powered by implicit neural representations.

It's like having more than five senses

This model transforms the captured data into a continuous and organic image of the environment. Different from traditional fragmented maps, the robot begins to recognize patterns and paths instinctively, as if it had real spatial awareness, even without direct vision and this significantly increases its safety and efficiency in dangerous areas, where errors are not an option.

During testing in a real forest park in the United States, the WildFision-equipped robot managed to cross areas of dense vegetation, unstable trails and open fields with a degree of confidence rarely seen, but with progress comes dilemmas.

When robberies acquire the ability to make autonomous decisions in unpredictable environments, serious ethical questions arise: How to guarantee that these systems operate safely in inhabited areas? And who will be responsible for eventual failures in critical missions?

Technology advances, but the debate needs to accompany, with robots learning to feel the world as we are facing a new era of artificial perception and perhaps for the first time instinct is being codified, but technology will respect the limits between autonomy and responsibility, we will continue accompanying to see where all this takes us.