DEPTRUM's sToF Deep Fusion Technology Opens a New Era of Robot Obstacle Avoidance

Obstacle avoidance is the core capability behind autonomous robot movement, directly determining whether a robot can complete its tasks safely and efficiently in complex environments. As the “eyes” of the avoidance system, depth sensors acquire precise obstacle distance information and provide key data for path planning.

Mainstream robot vacuum avoidance solutions today follow three technical paths: ToF (time of flight), line laser and binocular structured light. ToF, with its higher resolution, images small obstacles better, and its low computing cost supports higher frame rates.

However, iToF suffers from multipath effects, which introduce erroneous measurements into the depth information, disturb path planning and ultimately degrade avoidance. Multipath arises when signals reflect off different obstacles and reach the receiver along multiple paths with different delays, impairing ranging accuracy.

DEPTRUM's sToF deep fusion technology combines the high precision of structured light with the high resolution of ToF. Using actively projected coded spots from structured light, it effectively suppresses the multipath interference of conventional iToF and greatly improves ranging accuracy, achieving true complementarity. With millimeter-level obstacle recognition, sToF precisely senses thin, transparent or reflective objects such as wires, glass and mirrors, significantly reducing the risk of entanglement and collision.

Combining sToF with SLAM algorithms unifies navigation and avoidance in a single module, merging the avoidance and navigation modules of traditional solutions into one. Through DEPTRUM's proprietary mToF (modulated ToF) technology, laser light is modulated in space to achieve high-precision 3D sensing within 10 meters with a measurement error below 1%, greatly improving mid- and long-range mapping and localization.

With avoidance and mapping unified, the robot can output 3D obstacle point clouds through temporal fusion based on precise real-time 3D SLAM localization, far surpassing traditional single-frame depth avoidance. It improves overall avoidance efficiency and precision while reducing hardware space and cost, helping robot vacuum makers build more competitive products.