The spectacular performances at the 2025 World Robot Conference (WRC) were breathtaking, with humanoid robots capable of running, jumping, and even performing kung fu. While marveling at these impressive feats, we should also reflect on how these robots can truly integrate into our lives and become useful assistants.

The answer lies not in greater speed or strength, but in more precise perception and more natural interaction. Current humanoid robots are more like giants with powerful bodies but lacking vision. They cannot truly understand the three-dimensional world and cannot navigate complex environments as freely as humans.

Some netizens have noticed that most robots have a row of "small black holes" on their chests or foreheads. These are actually arrays of visual and spatial sensors, and the related technology is the spatial computing field that Xvisio Technology specializes in. They provide robots with "insight" and "thinking," allowing them to understand their surroundings in real time. This is a key step in giving robots a "soul."

We focus on three-dimensional "human-computer" interaction and put algorithms into the "eyes" of big machines. This is specifically reflected in the following three points:

01 Self-developed VSLAM builds 3D maps in real time

Allow the robot to complete millimeter-level positioning of "where am I and how far are you from me" in an instant;

02 Multimodal Fusion Interaction Engine

By combining vision, voice, and IMU signals into an ultra-low latency interactive link, the robot can raise its hand without spilling coffee.

03 Lightweight Inference Framework

A low-power chip can also identify in real time where people are and where their hands are reaching, so the robot won’t bump into you when handing you coffee.

04 "In other words, when humanoid robots collectively make heart gestures and become a trending topic, they are not comparing our outer shells, but the heartbeats hidden within our code."

The next generation of human-computer interaction technology we are developing is based on spatial computing and AI, making interactions as natural and intuitive as those between people. We believe that the best form of interaction is "unconscious" interaction—one so natural that you forget it's there.

Let's bring the engineers in front of the camera and answer five questions at once. -- Quick Q&A about the SeerSense® DS80 Perception Interaction Module

Netizen @Xiaobei:

At this World Robot Conference, humanoid robots gathered together to show off their skills. How do they "see" the world?

The DS80 relies on these "electronic eyes." It packs five major engines—binocular depth, Time of Flight (TOF), VSLAM, AI reasoning, and video encoding—into a compact 93g package. Plug it into a USB-C port and it gives the robot a third eye. In other words, many of the humanoid robots that flex their muscles live are rooted in spatial computing and perceptual interaction technologies.

Netizen @Doraemon:

Depth cameras are not uncommon, what makes DS80 different?

Adaptive environmental understanding, dual-mode deep learning engine for intelligent decision-making:

Passive binocular vision: Large field of view (110°), high resolution (640*480@60fps), 5.5-meter effective range (error <3%), excellent performance in bright light environments, suitable for open scene modeling and obstacle avoidance.

Active iTOF (Sony VGA): 940nm interference-resistant wavelength, ultra-high accuracy within 0.2-4 meters (error <1%), 30fps frame rate, and output of depth maps, point clouds, and IR images. Its accuracy far exceeds that of binoculars in low-light and indoor environments.Supports single-frequency (1.5m)/dual-frequency (4m) mode switching.

The dual engines can work simultaneously and seamlessly integrate with VSLAM to provide real-time environmental depth information, laying a solid foundation for navigation, obstacle avoidance, and 3D reconstruction.

The two engines can run simultaneously, allowing the robot to accurately shake hands in the exhibition hall and run and chase a ball in the sun.

UP host @VR Xiaofei:

SLAM sounds great, but will it be painful to develop?

Rich SDK: Supports Windows/Linux/Android, provides wrapper APIs such as ROS/ROS2, Unity, C/C++, Python, Java JNI, etc., and seamlessly connects to various development frameworks.

Highly expandable: USB Type-C connects to the host controller, and can be connected to sensors such as lidar through the expansion interface, supporting multi-machine cascading to meet complex system requirements.

Global Certifications: CE, FCC, RoHS, CB, FDA Class I laser safety certifications, suitable for global markets.

If you write three lines of code, the robot can build a map and locate itself at the same time, and can also automatically recognize the QR code to reset the "spatial anchor" - it's like giving the robot a "living map" that will never get lost.

Entrepreneur @Mr. Li:

I just want to make a human-computer interaction demo, but I don’t have a GPU. What should I do?

The key breakthrough is the independent hardware CNN dual engine! Run models locally without consuming host computing power. Simply drag and drop OpenVINO-trained models into the system for execution, starting at 30fps. Simply plug in and run AI.

Investor @Grace:

Besides humanoid robots, where else can this module be used?

All scenarios that require "understanding the three-dimensional world"——

Industry: AGV navigation, container volume measurement

Medical: Rehabilitation Robot Space Follow

Security: 3D facial recognition gate, area intrusion detection

Entertainment: Virtual Production Real-time Keying

In a word, "DS80 is a highly integrated spatial computing unit for the Metaverse and AIoT."

The end of the Robotics Conference marks the beginning of a new era. We firmly believe that after the hardware and computing power race, the next decade will be the decade of human-computer interaction.

Humanoid robots will move from being able to move to being able to understand; smart devices will move from passively receiving instructions to actively providing services; and the boundaries between the physical and digital worlds will become increasingly blurred. At the core of all this will be natural, seamless, and efficient human-computer interaction technology.

As a dedicated company, we are building this future. We not only provide technology and products, but also hope to become a thought leader and key enabler in the industry.

We welcome all individuals who are curious about the future and those seeking technological cooperation to join us in building this future of infinite possibilities.