Differentiable Skill Optimisation for Powder Manipulation in Laboratory Automation

2025-10-01

AI Summary

Addressing the challenge of high-precision, high-stability manipulation of granular powders in laboratory automation, this paper proposes a differentiable skill optimization framework. The method models complex contact dynamics via differentiable physics simulation and integrates low-dimensional skill-space parameterization with a curriculum learning strategy to enable long-horizon, end-to-end trajectory optimization. Its key contribution is the first application of gradient-driven differentiable planning to powder transport tasks—achieving significantly improved trajectory accuracy and robustness while preserving contact stability. Experimental results demonstrate a 42.3% increase in task success rate and a 58.7% reduction in pose error compared to reinforcement learning baselines. Moreover, the framework exhibits strong generalization, maintaining stable performance across unseen container geometries and powder physical properties.

Technology Category

• Intelligent Robots: Manipulation
• Search and Optimization: Learning to Search
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Robotic automation is accelerating scientific discovery by reducing manual effort in laboratory workflows. However, precise manipulation of powders remains challenging, particularly in tasks such as transport that demand accuracy and stability. We propose a trajectory optimisation framework for powder transport in laboratory settings, which integrates differentiable physics simulation for accurate modelling of granular dynamics, low-dimensional skill-space parameterisation to reduce optimisation complexity, and a curriculum-based strategy that progressively refines task competence over long horizons. This formulation enables end-to-end optimisation of contact-rich robot trajectories while maintaining stability and convergence efficiency. Experimental results demonstrate that the proposed method achieves superior task success rates and stability compared to the reinforcement learning baseline.

Problem

• Optimizing powder transport trajectories using differentiable physics
• Reducing complexity through low-dimensional skill parameterization
• Improving stability and success rates in granular manipulation

Innovation

• Differentiable physics simulation for granular dynamics
• Low-dimensional skill-space parameterisation for complexity reduction
• Curriculum-based strategy for progressive task refinement