BridgeDP at Robot Conference 2025: Three core technologies lay the foundation for general-purpose robot “cerebellum”
Three core technologies build the cornerstone of general-purpose robot “cerebellum”

Introduction
From August 8 to 12, the 2025 World Robot Conference (WRC) was held at the Beijing Beiren Yichuang International Exhibition Center. As a leading domestic developer of general-purpose robot “cerebellum” (motion control systems), BridgeDP showcased three major technological achievements. Its “precise, flexible, and efficient” cutting-edge solutions not only injected new momentum into general-purpose robot motion control but also provided key support for industry standardization and large-scale application, attracting wide attention from industrial production, automotive manufacturing, and home services sectors.
At the exhibition, BridgeDP founder Shang Yangxing gave the first public demonstration of the company’s upcoming general-purpose robot motion development platform. Notably, 16 robot manufacturers on site used BridgeDP’s motion control solutions, fully demonstrating the industry recognition and breadth of application of its technology.

BridgeDP founder Shang Yangxing demonstrates the general-purpose robot motion development platform
During the exhibition, Huang Xiangyue, Secretary of the Nanshan District Committee of Shenzhen, visited the BridgeDP booth for guidance, and gained in-depth understanding of the company’s innovative achievements and solutions in general-purpose robot motion control.

Core Technology 1: Omnidirectional locomotion control solution—equipping robots with mobility
BridgeDP’s omnidirectional locomotion control solution balances “realism” and “robustness” through an innovative algorithm architecture. It adapts to humanoid, quadruped, and wheeled-legged robots of various configurations, and learns multiple locomotion styles. It supports not only basic movements such as forward, backward, lateral movement, turning in place, walking, and running, but also complex scenarios like perception-enabled stair climbing. By optimizing the discriminator and policy algorithms, the robot’s movements become closer to the natural state of humans or animals while maintaining strong anti-interference capability.
Compared with traditional control schemes, this solution does not require finely designed reward functions. Through adversarial learning alone, the robot can autonomously master energy-efficient and stable locomotion patterns, greatly reducing the difficulty of algorithm development.

Quadruped robot dog with perception crossing obstacles
Core Technology 2: Whole-body imitation control solution—high-precision replication of complex human movements
BridgeDP’s whole-body imitation motion control solution, developed based on the Deepmimic algorithm, uses a two-stage approach of “imitation learning + reinforcement learning” to enable humanoid robots to transfer complex human movements with high precision. It supports complex actions such as dance and martial arts, and also supports customized motions.
The core logic of this technology is: first, human motion clips are captured by motion capture equipment, and then imitation learning is used to let the robot “replicate” the basic movement framework; afterwards, reinforcement learning optimizes physical feasibility to ensure stability and smoothness of the movements on the robot hardware.

Humanoid robot whole-body motion control
Core Technology 3: General-purpose robot motion development platform—lowering the barrier to motion development
BridgeDP founder Shang Yangxing gave the first public demonstration of the general-purpose robot motion development platform. As the core product embodying BridgeDP’s accumulated technical experience, the platform is already in internal use and will soon be officially launched to the market. The platform builds on the rich experience and industry insights gained from previously solving motion problems for clients with dedicated manpower. It integrates large-scale motion datasets, data preprocessing tools, automated model training, automated deployment testing, and continuous on-robot operation and maintenance. It greatly lowers the technical barrier to motion development and improves the development efficiency of robot motion control, providing key support for general-purpose robots to quickly adapt to multiple scenarios and achieve large-scale application. Developers can complete motion generation and debugging through low-code operations, unleashing their creativity in robot development.
Behind the technology: a hardcore team from “academic lab” to “industrial deployment”
BridgeDP was founded in May 2023. The core technical team has an average age of only 26 and brings together robotics talents from top domestic and international universities such as HUST, SUSTech, CMU, and Beihang. Founder Shang Yangxing has been deeply involved in robot development since his undergraduate years. With a full-chain capability spanning “academics + engineering + business,” the team achieved the leap from technological breakthroughs to industrial deployment in just two years.
The company aims to “fill the gap in the third-party motion control market” and directly addresses three major industry pain points: difficult physical transfer, low development efficiency, and poor software versatility. By using technologies such as Sim2Real training, reinforcement learning, and deep imitation, it calls on basic motion control modules on demand, reducing development costs and improving efficiency. A set of motions can be developed in as little as 3 days, with motion accuracy raised to 92% and transfer success rate reaching 80%. It has become one of the few domestic enterprises focused on providing general-purpose motion control solutions. As of Q2 2025, its motion control solutions have been deployed in more than 40 robot manufacturers.
BridgeDP’s ultimate goal is to make robot motion control extremely simple and convenient. To this end, BridgeDP is committed to the technology research and algorithm iteration of a general-purpose robot motion control platform to meet customers’ motion development needs. Moreover, based on its judgment of the future, BridgeDP’s business model is attempting to shift from project-based delivery to a model of license + service fees + subscription. BridgeDP hopes that robot motion control technology can be general-purpose, efficient, and accessible, so that robots entering thousands of industries will no longer be a distant vision.






界面新闻
深视新闻
深圳特区报
财新
36氪
钛媒体
中国新闻周刊
蓝鲸新闻
Z Finance
投资家网
雷峰网
亿欧
环球网
南方日报
极客公园



