Mana: Dexterous Manipulation of Articulated Tools

Noosaga Placements

• Learning-Based Manipulationframework95%
  The paper explicitly uses learning-based manipulation, specifically RL and diffusion policies, to generate and execute manipulation trajectories.
  Mana employs a coarse-to-fine pipeline...through motion planning and reinforcement learningWe train a point-cloud-conditioned transformer diffusion policy
• Robot Manipulationsubfield95%
  The paper directly addresses dexterous manipulation of articulated tools, which is a core problem in robot manipulation.
  Articulated tool manipulation remains a major challenge in dexterous roboticsWe present Mana (Manipulation Animator), a general sim-to-real framework for dexterous manipulation
• Robot Learningsubfield90%
  The framework uses reinforcement learning and diffusion policies to generate and learn manipulation trajectories.
  motion planning and reinforcement learning to generate manipulation trajectoriesWe train a point-cloud-conditioned transformer diffusion policy
• Deep Reinforcement Learningframework85%
  The framework uses reinforcement learning for short, contact-rich manipulation phases, which is a core part of the deep RL framework.
  reinforcement learning is used only for the short, contact-rich phases where delicate position-force coordination is requiredWe use the same RL formulation described in the next section for challenging cases
• Roboticssubfield70%
  The work bridges robotics and AI through learning-based methods and sim-to-real transfer.
  sim-to-real reinforcement learning approach is also insufficient on its owntrain a point-cloud-conditioned transformer diffusion policy
• Optimization-Based Motion Planningframework70%
  Motion planning is used for the pre-grasping phase to generate collision-free trajectories.
  We instead implement a GPU-accelerated RRTConnect algorithm with path smoothing to plan trajectories

Abstract

Articulated tool manipulation remains a major challenge in dexterous robotics due to the need to coordinate internal degrees of freedom and contact-rich interactions. While prior work has largely focused on rigid objects, articulated tool use remains underexplored because of its physical complexity and the difficulty of learning functional grasping and manipulation policies. We present Mana (Manipulation Animator), a general sim-to-real framework that reinterprets dexterous manipulation as an animation problem. Inspired by computer animation, Mana employs a coarse-to-fine pipeline that transforms procedurally-generated grasp keyframes into manipulation trajectories through motion planning and reinforcement learning. The data generation process is largely automatic, requiring only a few mouse clicks to specify functional affordances (<1 minute per tool). Across four articulated tools spanning different scales and joint types, Mana achieves zero-shot sim-to-real transfer for both grasping and in-hand manipulation, demonstrating a scalable approach to dexterous articulated tool use.

Paper Context