BridgeDP raises hundreds of millions in new funding, accelerating general robot OS

Recently, BridgeDP Robotics announced the completion of its Pre-A+++ round of financing. This round was led by China Mobile Chain Fund, with follow-on investments from existing shareholders Fosun RZ Capital and Shenzhen Capital Group (SCGC), reaching a scale of hundreds of millions of RMB.

Along with this round of financing, BridgeDP has further clarified its long-term strategic direction: to establish a "General-Purpose Robot Operating System" with motion control as its foundation. This also marks our transition and upgrade from a developer of motion control solutions to a builder of general-purpose robot industry infrastructure and ecosystem.

Within half a year, BridgeDP has successively secured investments from industrial capital and state-owned funds. This is not only a recognition of our past achievements, but also a vote of confidence and trust in our new positioning and new direction.

This round of financing will be used to accelerate the implementation of this strategy, driving our layout in product iteration and algorithm R&D into the next stage.

RoboCraft AI

General Robot Motion Capability Development Platform

We launched RoboCraft AI in June this year. It consolidates BridgeDP's past experience in the development and deployment of over 50 different robot models, standardizing underlying motion control technologies, model algorithms, and Sim2Real engineering experiences into standard software development workflows.

We hope that developers can quickly build world-class motion capabilities for robots through RoboCraft AI.

On one hand, for multiple mainstream mature models, the platform supports rapid development of scene applications and physical deployment.

RoboCraft AI is already compatible with models like Unitree G1, Zhiyuan X2, Team-Z PM01, and Accelerated Evolution T1. For rental providers, integrators, and content teams who have purchased these models, we offer an "out-of-the-box" application development toolchain, bringing "motion customization, application orchestration, and physical deployment" into a single workspace. Specifically, users can:

• Use the platform's rich asset library of high-quality motions such as walking, running, and dancing;

• Add custom actions through "video extraction" or "uploading BVH files";

• Perform seamless application orchestration of multiple motions on the timeline via simple drag-and-drop, and save them for long-term management;

• Deploy applications to physical robots with one click and enable multi-robot fleet control

During one month of closed beta testing, RoboCraft AI has established connections and partnerships with dozens of robot rental and performance enterprises nationwide. Multiple clients have independently completed the deployment process, significantly boosting development efficiency in commercial performance scenarios through the platform.

On the other hand, the platform also supports the integration and adaptation of self-developed models from robot manufacturers.

For self-developed robot hardware companies, RoboCraft AI provides a standardized onboarding process, transforming complex tasks that were previously costly and heavily reliant on expert experience into reusable platform automated functions, significantly shortening the cycle of deploying motion capabilities on new models.

RoboCraft AI has developed features ranging from AI verification and auto-repair of URDFs, to hardware safety checks, and automatic generation of training parameters. It attempts to preemptively resolve hardware vulnerabilities in Sim2Real engineering, improving subsequent deployment success rates and efficiency.

We have reached deep collaborations with multiple complete robot manufacturers and joint module suppliers, successfully running 0-to-1 adaptation for their new models, helping partners acquire near-mature basic motion capabilities and secondary development space.

Currently, RoboCraft AI is fully open for registration. Visit the official website robocraftai.cn to unlock high-quality built-in motion libraries and the full application development toolchain. Industry partners are welcome to test and provide feedback!

General Motion Control Model

Powerful Foundation for Motion Capabilities

The cross-body deployment and motion generalization capabilities exhibited by RoboCraft AI are fundamentally rooted in BridgeDP's general motion control model.

Our algorithm team has integrated motion data and deployment experience accumulated from serving different robots in the past into the R&D of our general motion tracker. This achievement is based on our self-developed general motion tracking model architecture, combined with thousands of hours of self-collected cross-body whole-body motion data for training. Currently, it stably supports high-frequency actions such as walking, running, jumping, dancing, and martial arts, as well as complex dance choreographies not previously adapted individually, demonstrating excellent action generalization.

With the capacity expansion of BridgeDP's cross-body whole-body motion data factory, the model's action coverage, generalization of complex actions, and physical deployment stability will continue to improve in the future.

Furthermore, the general motion tracker is not merely a single-point action model, but can also interface with upper-level motion planner outputs. Currently, we are compatible with the outputs of both explicit and implicit walking planners, converting upper-level planning results into executable, deployable bottom-level whole-body motion control capabilities.

Subsequently, based on this foundation, our interface capabilities will gradually expand to more complex tasks such as navigation and manipulation, allowing the motion control model to extend beyond mere "action execution" and become the core foundational motion capability layer in the "General-Purpose Robot Operating System" BridgeDP aims to build.

Cross-Body Whole-Body Motion Data Factory

The Core Engine for Continuous Capability Evolution

In early June this year, our self-built cross-body whole-body motion data factory went into operation. It has currently produced its first batch of thousands of hours of valid training data that has passed full-process quality control. The production line has transitioned from initial workflow proof to a new stage of capacity scaling.

Regarding data planning, we adopt a dual-track production mode of "proactive coverage + model capability demand feedback":

• For proactive data collection, the factory coordinates a professional motion design team with an AI-native management platform, continuously expanding the motion library across dimensions such as locomotion, pose transitions, whole-body interactions, and fall recovery to ensure data diversity and validity;

• For model capability demand-driven data collection, feedback on training failures, cross-body transfer degradation, dynamics fallback, and low-yield samples is optimized and routed to the next round of collection and retargeting, ensuring data production serves actual gaps in model iteration.

At the same time, the data factory is equipped with a self-developed cross-body motion retargeting algorithm, which quickly adapts and distributes collected data to different robot models, achieving efficient reuse of high-quality motion data across multiple robot platforms, significantly improving data production and application efficiency, and further accelerating the training, validation, and capability evolution of cross-body motion control models.

Once again, we would like to express our gratitude to all investment institutions and industry partners for their long-term recognition and trust in BridgeDP within the field of motion control.

Moving forward, while continuing to deepen our core business advantages, we will increase investment in our data factory, algorithm R&D, and product iterations to lay a solid foundation for completing the construction of the "General-Purpose Robot Operating System".

We hope to provide the industry with a foundational infrastructure for general-purpose robot application development, allowing more people to participate in the future construction of the robot ecosystem. We believe this is a powerful path to boost industry efficiency and drive industrial innovation.

As Shang Yangxing, founder of BridgeDP, said: "BridgeDP's mission is to enable intelligence to truly enter and transform the physical world. To achieve this, we cannot rely on just a few companies, nor on a few genius engineers. We need to allow more developers, hardware teams, scenario experts, and entrepreneurs to participate in building physical AI."