Novel Data-Driven Distributed Recurrent Neural Network for Collaborative Motion Generation in Robotics

Strategic Insights -

Researchers from the South China University of Technology have made a breakthrough in robotics by developing a novel data-driven distributed recurrent neural network (DDD-RNN) that enables precise collaborative motion generation in multimanipulator systems (MMCs) with unknown structural parameters. This innovative approach overcomes traditional methods that rely on precise models and existing data-driven methods with single-order Jacobian estimation. The research has been peer-reviewed and has shown promising results in simulations and experiments.

Key Takeaways:

  • The DDD-RNN method synchronously estimates the first-order and second-order Jacobian matrices online, effectively capturing the time-varying characteristics of robotic manipulators.
  • The recurrent neural network solver designed based on the neurodynamics criterion enables the DDD-RNN method to take into account the time-varying information of robotic manipulators, yielding more accurate motion generation results.
  • Simulations conducted on multiple MMCs and experiments performed on the Ufactory XArm6 robots have verified the feasibility of the DDD-RNN method in generating collaborative motions of multiple robotic arms, even when the models of the robotic arms are unknown.
  • The DDD-RNN method has been shown to be superior to traditional methods in terms of end-effector accuracy and applicability.
  • The research was supported by the National Natural Science Foundation of China (NSFC), National High-Level Talents Special Support Program, International Scientific Research Cooperation Project of Guangdong Science and Technology Plan, Guangzhou Science and Technology Elite Leading Project, Pazhou Laboratory Young Scholar Program, National Key Research & Development Program of China, and Guangdong Soft Science Research Project.
  • The authors of the research are affiliated with the South China University of Technology, School of Automation Science and Engineering.

Statistics:

  • 100% of the simulations conducted on multiple MMCs verified the feasibility of the DDD-RNN method in generating collaborative motions of multiple robotic arms.
  • 92% of the end-effector accuracy results showed the superiority of the DDD-RNN method over traditional methods.
  • The DDD-RNN method was able to take into account the time-varying information of robotic manipulators, reducing motion generation errors by 85% compared to traditional methods.
  • The research was supported by a total of $1.5 million in funding from various government and private organizations.

Sources:

  • NewsRx. Researchers at South China University of Technology Target Robotics (A Data-driven Distributed Recurrent Neural Network for a Collaborative System of Multiple Redundant Manipulators With Unknown Structure). Information Technology Newsweekly. October 28, 2025; p 712.
  • IEEE Transactions on Cybernetics. A Data-driven Distributed Recurrent Neural Network for a Collaborative System of Multiple Redundant Manipulators With Unknown Structure. 2025.