Author | Zhang Lianyi

Editor | Qiu Kaijun

'Hello Robot,' this event marks not merely a product launch but the dawn of a new era in robot technology.

On January 3, RoboSense hosted the 'Hello Robot' 2025 AI Robot Global Launch Event, setting the stage for high expectations.

At the event, RoboSense CEO Qiu Chunchao outlined the company's comprehensive strategy for a robot technology platform. Building on this strategy, RoboSense unveiled a suite of new products, including a fully assembled humanoid robot, underscoring its commitment to the robot business.

With these new products, RoboSense stands on the cusp of entering the second decade of the intelligent robot era. Despite persistent skepticism about LiDAR, the company continues to thrive.

01 A Decade of Growth: From LiDAR to Intelligent Robots

RoboSense's foray into robots is not new.

As early as the end of November 2024, during RoboSense's third-quarter earnings call, Qiu Chunchao had already signaled the company's preparations for a robot business.

At the event, Qiu Chunchao elaborated on why RoboSense is well-positioned for this endeavor.

RoboSense CEO Qiu Chunchao

Contrary to the perception of RoboSense solely as a LiDAR supplier, Qiu Chunchao emphasized that the company's aspiration has always been to become a global leader in robot technology platforms. 'Our name and logo clearly reflect our original dream.'

In a narrow sense, 'RoboSense' represents 'robot + perception,' while in a broader context, it signifies 'robot + cognition,' i.e., 'AI + robot.'

When RoboSense was founded a decade ago, it developed its first product to address the perception challenges faced by robots in task execution, eventually becoming a leader in the automotive LiDAR industry.

Data indicates that by the end of the third quarter of 2024, RoboSense had achieved mass production and delivery for 31 vehicle models. As of November 29, 2024, the company had secured mass production orders for 92 vehicle models from 28 automakers and tier-one suppliers.

Selected RoboSense Designated Vehicle Models

While the automotive business has been growing, gross margins have remained relatively low. Financial reports show that as of the end of the third quarter of 2024, the company's gross margin was 17.5%, with a gross margin of 14.1% for LiDAR products used in ADAS applications.

According to Qiu Chunchao, the robot market, due to its smaller size compared to the automotive market, has traditionally maintained gross margins between 35% and 50%.

At the same time, Qiu Chunchao believes that the growth rate in the robot sector mirrors the explosive development seen in the automotive industry from 2021 to 2023. With the onset of a second growth curve, shipments in the robot sector are expected to exceed six figures by 2025.

This is a crucial reason why RoboSense is firmly committed to the robot market.

What many may not know is that when RoboSense first entered the LiDAR industry, it focused on static surveying and industrial robots, accumulating a certain number of users and shipments.

According to Qiu Chunchao, RoboSense has provided incremental components and solutions for over 2,600 robot and other industrial clients across various scenarios and tasks.

With growing mass production experience, RoboSense has established a presence in hardware, chips, and AI technology, providing the foundational capabilities for becoming an AI-driven robot technology platform company.

In terms of hardware, in its first decade, RoboSense focused on LiDAR research and development, accumulating core competencies in system electronics, power electronics, and optical structures.

In terms of chips, RoboSense has achieved full chip integration across the four subsystems of scanning, emission, reception, and processing within the LiDAR system.

In AI, RoboSense has not only built AI infrastructure such as supercomputing centers, AI models, and scenario data but also developed large-scale data closed-loop toolchains, continuously enhancing AI capabilities through data-driven methods.

In Qiu Chunchao's view, over the past decade, among the myriad stars in the RoboSense constellation, LiDAR has been the brightest. However, in the next decade, with the explosion of the intelligent robot industry, RoboSense is expanding into the vast 'galaxy' of intelligent robots.

02 Humanoid Robot Debut: Focusing on Vision, Touch, and Joints

With a clear goal in mind, how does one achieve it?

RoboSense's approach is to develop versatile robot mobility and manipulation solutions based on complete machine development and introduce incremental components for intelligent robots suitable for various scenarios. This will lower the barriers to entry, shorten development cycles, accelerate the commercialization of the robot industry, and drive the evolution towards embodied intelligent robots with general AI capabilities.

To this end, at the event, RoboSense showcased its comprehensive capabilities by unveiling a humanoid robot as a development platform for versatile robot components, focusing on vision, touch, and joint incremental component technologies.

Among these, the most significant product is the new vision category, the Active Camera solution.

According to Qiu Chunchao, the Active Camera is an intelligent ecosystem comprising sensor hardware, computing cores, and AI algorithms, aiming to become the universal eye of intelligent robots.

At the hardware level, unlike traditional passive cameras, the Active Camera integrates LiDAR digital signals with camera information, enabling robots to not only recognize precise 3D environmental distance information but also perceive rich visual semantic information, overcoming environmental interference such as drastic changes in brightness.

In terms of deployment and application, robot developers can directly utilize the Active Camera's mapping, positioning, obstacle avoidance, and other capabilities through an SDK, saving time spent on sensor driver development, data calibration, data fusion, and avoiding system inefficiencies caused by stacking multiple sensors.

For touch, RoboSense has developed a low-cost force sensor, FS-3D, suitable for end-effector motion control in legged robots, enhancing the precision of bipedal robot motion control.

For joints, a high-power-density linear motor, LA-8000, has been developed. With a peak thrust of up to 13,000N, it provides leg strength for humanoid robots in scenarios such as carrying heavy loads, running, and jumping.

Additionally, RoboSense has introduced the DC-G1, a highly integrated, compact, high-performance, and low-power robot domain controller, serving as the central computing platform for intelligent robots. It supports various real-time perception models, multimodal large models, embodied intelligent operation models, and motion control algorithms, assisting developers in achieving intelligent product upgrades.

Of course, it's not enough to just 'see' and 'run'; robots must also be able to 'manipulate'.

Last November, RoboSense showcased an 8-degree-of-freedom robotic component called the 'Dexterous Hand,' which utilized RoboSense's self-developed 3D force sensor technology.

Now, RoboSense has released the second-generation Dexterous Hand, Papert 2.0, which Qiu Chunchao believes to be among the top performers in the market in terms of performance and configuration.

Designed to mimic human hands, Papert 2.0 features 20 degrees of freedom and a maximum load capacity of 5kg, with 14 force sensors distributed across the fingertips, finger pads, and palm.

Paired with a robotic arm and its control system, Papert 2.0 can replicate fine human hand movements and operations. It can use its four-degree-of-freedom index finger to operate an electric screwdriver or gently pick up an egg with precise force...

This 'hand-eye coordination' combination is defined by RoboSense as the robot manipulation solution.

For the robot's mobility solution, RoboSense, leveraging its experience in intelligent vehicles as mobile robot platforms, will provide the industry with high-speed or low-speed autonomous driving solutions for various mobile robots.

From head to toe, RoboSense provides corresponding solutions.

Regarding how to conduct this business, RoboSense has already provided a solution.

As a technology platform for upstream robot clients, RoboSense offers a full stack of technical capabilities ranging from software algorithms to chips and even motors, with great flexibility.

For example, for companies with comprehensive robot solutions, RoboSense only provides standardized hardware support. For companies lacking relevant capabilities, RoboSense can provide a full suite of solutions, including software, hardware, and development toolchains. This positioning can be compared to that of a Tier 1 supplier like Bosch.

03 Comprehensive Upgrade of the 'Old Business' - LiDAR

While advancing its robot business, RoboSense is obviously not abandoning LiDAR and is even upgrading it comprehensively.

'A decade ago, the representative of intelligent robots was the automobile.' In Qiu Chunchao's view, after a decade of effort, LiDAR has fully entered the digital era, marking the beginning of a new chapter.

The first act of this new chapter is the upgrade of automotive LiDAR.

At the event, RoboSense unveiled the EM4, the world's first 'thousand-line' ultra-long-range digital LiDAR, the first product under its new digital EM platform for the automotive sector.

According to Qiu Chunchao, the EM4 integrates multiple technologies such as a digital architecture, crosstalk elimination, full-condition optoelectronic signal processing, and lossless data compression. It boasts up to 1080 lines, a maximum ranging distance of 600 meters, and an imaging capability of 25.92 million points per second, ushering in the 'thousand-line' era for automotive LiDAR.

The EM4 provides vehicles with 1080P high-definition 3D perception capabilities, enhancing the performance of advanced intelligent driving and fully autonomous driving.

For the intelligent robot market, RoboSense launched two new digital LiDAR products: E1R and Airy.

Among them, the E1R is a fully solid-state digital LiDAR equipped with a digital SPAD-SoC chip and a 2D VCSEL chip. With a compact light-emitting window and a super-wide field of view of 120°×90°, it can detect object size, contour, distance, and other information in real-time.

Designed based on an automotive-grade platform, it is suitable for various types of mobile robots in industries, commerce, and embodied intelligence, navigating through different lighting conditions to meet the needs of fine perception at close range and wide-range environmental detection, enhancing operational efficiency and safety of intelligent robots.

Qiu Chunchao stated that E1R has gradually started mass production and delivery and will be one of RoboSense's main products in the robot sector in 2025.

The other product, Airy, is a 192-line hemispherical digital LiDAR. With a size comparable to a ping-pong ball, it provides a super-wide hemispherical FOV of 360° horizontally and 90° vertically, covering a diameter of 120m, with a detection accuracy of ±1cm and a point output rate of 1.72 million points per second. It can detect object size, contour, distance, and other information in real-time, suitable for robot applications requiring omnidirectional detection.

In Qiu Chunchao's view, just as film cameras evolved into digital cameras, LiDAR has ushered in a new era of full digitization. From now on, LiDAR architectures will mature and stabilize, with products from different technical routes undergoing digital upgrades, accelerating their penetration and popularization in the automotive, robot, and drone markets.

In the future, RoboSense will also introduce a series of digital LiDAR products tailored for different robot scenarios.

With the release of a series of products, RoboSense is poised to enter the second decade of the intelligent robot era, unlocking the second curve of growth.

END-