We have always been thinking: Can robot motion development be as simple as cutting videos and editing documents? Faced with humanoid robots of different brands and configurations, can there exist a unified, complete development toolchain and deployment process in the industry, building continuously evolving motion capabilities for robots?

The reality is that both hardware manufacturers and buyers are facing different industry challenges.

For hardware manufacturers, developing motion control algorithms from scratch is a costly and hard battle. In the traditional model, motion control development is a black hole that tests "composite capabilities" and "engineering experience", requiring the team to be proficient in dynamics control, reinforcement learning, simulation modeling, and Sim2Real deployment. The implementation of actions heavily relies on the team's hands-on experience and troubleshooting capabilities. This is a capital-intensive, long-cycle undertaking, requiring a lot of manpower and time costs.

For buyers downstream of the industry chain, due to the closed and fragmented upstream ecosystem, in order to make robots meet different user needs in real scenes, buyers need to have二次 development capabilities. However, in the past, this was highly dependent on experience, requiring engineering and algorithm capabilities, which is unfriendly to most robot users.

At this stage, what the industry needs is not just a breakthrough in single-point algorithms, but a unified, reusable, and standardized motion development platform.

RoboCraft AI:

Universal Robot Motion Development Platform

Based on these industry pain points, BridgeDP has launched RoboCraft AI—a universal robot motion development platform.

It condenses BridgeDP's algorithm accumulation and Sim2Real engineering experience across more than 50 robot projects with different brands and configurations, transforming expert-level technology into unified, reusable platform capabilities and standardized development flows.

We hope developers can leverage this platform to quickly build first-class motion capabilities for robots.

Below is the RoboCraft AI development capability demonstration:

Facing the dual pain points of manufacturers and downstream developers, RoboCraft AI solves them from two dimensions.

Path 1:

Self-developed Robot 0→1 Motion Control Access

For self-developed robot enterprises, RoboCraft AI provides a fast, unified integration process, gradually standardizing stages that previously relied heavily on expert experience:

• Intelligent URDF Model Validation: Robot structural issues often only expose themselves when running on real hardware, bringing safety hazards. The platform can detect, notify, and fix issues early in the URDF phase.

• Hardware Safety Process and Validation: Non-standard hardware issues are widespread. BridgeDP has transformed experience debugging various robots into standardized safety check processes.

• Automated Parameter Tuning in Training: Motion control model training requires repeated parameter tuning for different actions and environments, which is extremely costly. The platform achieves automated tuning, significantly reducing manual intervention.

  
  

Once integrated, self-developed robots can obtain basic motion control capabilities and development environments consistent with mature models, significantly shortening the cycle from hardware readiness to sustainable development.

Beta Support Scope: Integration process for self-developed robots is open for application. The specific models and cycles are subject to review and evaluation.

Path 2:

Action Choreography & Deployment for Mature Models

For users who have purchased mature models (lessors, integrators, content creators, research teams, etc.), RoboCraft AI provides an out-of-the-box action development toolchain, eliminating the need for repeated integration and training.

Action Acquisition & Choreography

• Support uploading MP4 videos to extract actions, and importing BVH action files;

• Visual timeline editing, low-code completion of action combinations and commercial show configurations;

• Built-in action asset library management, supporting accumulation and reuse within the team.

Multi-Robot Deployment & Group Control

• Batch deployment of robots, one-click application delivery;

• Multi-machine cooperative startup and standby.

Team Collaboration & Asset Management

• Unified management and sharing of action assets, app templates, and robot configurations;

• Assets developed by members can be stored with one click, avoiding duplicate development.

Beta-Compatible Models: Unitree G1, Zhiyuan X2; more brands and models are being continuously expanded.

Beta Application Is Officially Open

During the beta period, we will provide the following benefits:

• Free access and deployment of multiple popular dance applications for Unitree G1, Zhiyuan X2, and other models;

• Unlimited use of custom action customization features.

We especially welcome the following groups to apply for beta:

• Robot R&D enterprises that want to quickly obtain mature motion capabilities;

• Robot lessors and system integrators who want to quickly deploy robot commercial shows, cultural tourism, and educational exhibitions;

• Universities and research institutions wishing to carry out robot research and teaching;

• A new generation of developers who want to explore robot content creation, action development, and application innovation.

Scan to Apply:

Once approved, users can register, log in, and create organizations by themselves. We will also invite users to join the RoboCraft AI Core Beta Community to jointly promote product iteration with the BridgeDP team.