New Low-Power Vision System Revolutionizes Robot Navigation

A groundbreaking low-power robotics system, LENS, is poised to transform how robots perceive and navigate their environments. This system, which merges a brainlike sensor, a chip, and a tiny AI model, offers a significant leap forward in energy efficiency and practicality for a wide range of robotic applications.

LENS, short for Low-Energy Navigation System, is designed to be highly energy-efficient. It uses only 10 percent of the energy required by conventional location systems, as reported in the June issue of *Science Robotics*. This efficiency is crucial for robots involved in space and undersea exploration, as well as for drones and microrobots used in medical procedures.

According to Yulia Sandamirskaya, a roboticist at Zurich University of Applied Sciences who was not involved in the study, such a low-power 'eye' could be extremely useful. "It is, frankly, insane that we got used to using cameras for robots," she says, highlighting the energy-intensive nature of traditional camera-based systems.

The key components of LENS are a sensor and a chip called Speck, developed by the company SynSense. Speck's visual sensor operates more like the human eye, detecting changes in the environment rather than capturing everything in the visual field. Adam Hines, a bioroboticist at Queensland University of Technology in Brisbane, Australia, explains, "Each pixel of Speck’s sensor only wakes up when it detects a change in brightness in the environment, so it tends to capture important structures, like edges."

This approach is much more efficient than traditional cameras, which capture a vast amount of data, even if nothing changes. Mainstream AI models excel at processing this data, but the combination of camera and AI consumes a significant amount of power. Determining location can use up to a third of a mobile robot’s battery.

In contrast, the human eye and brain work together to detect and process changes efficiently. The LENS system emulates this by using a neuromorphic computing approach, where the sensor and chip work together with an AI model to process environmental data. The AI model, developed by Hines’ team, learns to recognize places by analyzing edges and other key visual information from the sensor.

The system's efficiency is a game-changer for robotics. "Radically new, power-efficient solutions for place recognition are needed, like LENS," Sandamirskaya emphasizes. This system could enable robots to operate for longer periods without recharging, making them more suitable for long-duration missions in challenging environments.

For example, a hexapod robot using LENS can navigate a path in a lush garden setting, determining its location with minimal energy consumption. The system could also be applied to drones, microrobots, and other devices where power efficiency is critical.

The potential applications of LENS are vast, from space exploration to medical diagnostics. By reducing energy consumption and improving navigation capabilities, this innovative system could usher in a new era of robotic technology.