RoboSense Launches AC2 ‘Robot Operation Eye’ at IROS 2025 with Millimeter-Level Precision

RoboSense Launches AC2 ‘Robot Operation Eye’ at IROS 2025 with Millimeter-Level Precision

πŸ‘€ admin πŸ“… Jul 17, 2026 πŸ• 2 min read

October 21, 2025 β€” Hangzhou, China β€” At the International Conference on Intelligent Robots and Systems (IROS 2025), RoboSense Technology (RoboSense) officially launched the AC2, the latest product in its Active Camera series. Positioned as the “Robot Operation Eye,” the AC2 is the industry’s first dToF + RGB stereo + IMU integrated super sensor system, designed to deliver millimeter-level fusion perception for robotic manipulation tasks.

Breaking the Precision Barrier

One of the biggest limitations in robotic manipulation has been insufficient perception precision. Most robots today can only perform basic operations because their 3D vision sensors cannot provide the accuracy needed for fine manipulation. The AC2 addresses this with breakthrough specifications:

  • Β±5mm precision within 8 meters β€” achieved through a redesigned full-solid-state area dToF depth perception system that overcomes the industry-wide problem of accuracy degradation with distance
  • Enhanced near-field precision β€” the integrated stereo vision system further strengthens depth accuracy within 1 meter, critical for close-up manipulation
  • 1600Γ—1400 high-resolution RGB camera β€” supplements depth information with ultra-clear color image data for richer fusion perception

Hardware-Level Fusion and Synchronization

The AC2 is not simply a combination of sensors β€” it achieves true hardware-level data fusion through RoboSense’s chip-level hardware synchronization control technology. This ensures:

  • 1ms synchronization precision between depth and image information across most scenarios
  • Flexible output of fused or independent depth, image, and motion posture data
  • Consistent, clean perception output across the entire field of view

The sensor provides a 120°×90Β° ultra-wide field of view, allowing robots to see more operational targets at once β€” eliminating the need for frequent head movements that introduce additional positioning errors.

Compact and Reliable

Despite its powerful capabilities, the AC2 measures only 100mm Γ— 30mm Γ— 45mm (LΓ—HΓ—D), making it flexible enough to mount on humanoid robot faces, robot dog heads, and robotic arm endpoints where space is extremely limited.

Based on the dToF depth sensing approach, the AC2 maintains stable performance in challenging lighting conditions β€” low light, strong light, and alternating bright-dark environments. Robots equipped with the AC2 can:

  • Perceive and manipulate small objects like toothbrushes and clothes hangers
  • Handle high-reflectivity surfaces like glass and metal
  • Maintain undistorted fusion perception during large robot movements

AI-Ready Ecosystem

Alongside the AC2 hardware, RoboSense updated its AI-Ready ecosystem with new open-source algorithms including posture estimation and human skeleton recognition β€” specifically adapted for AC2 use cases. The company also launched the AC1 earlier, which achieved 600% of traditional 3D camera ranging capability with a maximum range of 70 meters and 3cm consistent precision.

“Embodied intelligent robots should not be limited to understanding through human eyes,” said Professor Zhu Yanhe from Harbin Institute of Technology. “The Active Camera is essentially a ‘Super Eye’ that fuses cameras, LiDAR, and multiple sensors β€” walking at the forefront of innovation.”

Leave a Reply

Your email address will not be published. Required fields are marked *