This post is also available in:
עברית (Hebrew)
Most robots rely heavily on cameras and external tracking systems to understand their surroundings and control movement. But in cluttered, dark, or visually degraded environments, these systems can become unreliable. Soft robots face an additional challenge: because their bodies constantly bend and deform, their sensors often struggle to distinguish between self-generated movement and external contact.
Researchers have now developed a soft robotic system designed to solve this problem using a form of internal body awareness similar to human proprioception—the “sixth sense” that allows people to understand body position and movement without looking. According to Interesting Engineering, instead of depending on cameras, the robot uses internal sensing and predictive modeling to interpret how its body should move and compare that expectation with real-world feedback.
At the center of the system is what researchers describe as an “expected perception” framework. The robot predicts the outcome of its own movements and continuously checks those predictions against data collected from flexible liquid-metal sensors embedded throughout its structure. When differences appear between expected and actual movement, the robot interprets them as signs of external interaction, such as touching an obstacle or encountering resistance.
This approach allows the robot to navigate using touch and deformation awareness alone. In testing, a flexible robot successfully moved through a maze without any camera input, relying entirely on internal sensing to detect walls and adjust direction. Researchers also demonstrated the system in guided motion tasks, where the robot learned and reproduced complex movements with high precision.
The sensing system proved capable of detecting contact in less than half a second and identifying the direction of applied forces with relatively high accuracy, even in dynamic conditions. Because the framework is based on body awareness rather than visual data, it may be particularly useful in environments where cameras are limited by lighting, visibility, or physical obstructions.
From a defense and security perspective, touch-aware robotics could support operations in underground tunnels, underwater environments, smoke-filled spaces, or other areas where optical systems are degraded. Soft robots that navigate through physical interaction rather than visual mapping may also reduce dependence on vulnerable external sensing systems.
The research reflects a broader shift in robotics toward biologically inspired sensing, where machines increasingly combine physical adaptability with more human-like environmental awareness.
The research was published here.
