Technology Category
Intelligent Robots: ManipulationMachine Learning: Imitation Learning & Inverse Reinforcement LearningSearch and Optimization: Sampling/Simulation-based Search
Towards Automated Chicken Deboning via Learning-based Dynamically-Adaptive 6-DoF Multi-Material Cutting
🤖 AI Summary
Automated deboning of chicken shoulders poses a significant challenge due to the need for precise six-degree-of-freedom cutting through multi-material, highly deformable joints—contact with bone incurs severe food safety risks. To address this, we propose a force-feedback-driven residual reinforcement learning framework that integrates discretized force sensing with domain randomization to enable closed-loop, dynamic trajectory adjustment. We further develop an open-source, multi-material cutting simulator and a reproducible physical testbed, achieving— for the first time—zero-shot sim-to-real transfer to real chicken shoulders. Without any pretraining on real-world data, our system fully automates the deboning task: bone contact rate is substantially reduced, and success rate improves by up to 4× over open-loop baselines. This work demonstrates the critical role and practical viability of force-guided adaptive control in complex biological tissue manipulation.
Technology Category
Application Category
📝 Abstract
Automating chicken shoulder deboning requires precise 6-DoF cutting through a partially occluded, deformable, multi-material joint, since contact with the bones presents serious health and safety risks. Our work makes both systems-level and algorithmic contributions to train and deploy a reactive force-feedback cutting policy that dynamically adapts a nominal trajectory and enables full 6-DoF knife control to traverse the narrow joint gap while avoiding contact with the bones. First, we introduce an open-source custom-built simulator for multi-material cutting that models coupling, fracture, and cutting forces, and supports reinforcement learning, enabling efficient training and rapid prototyping. Second, we design a reusable physical testbed to emulate the chicken shoulder: two rigid "bone" spheres with controllable pose embedded in a softer block, enabling rigorous and repeatable evaluation while preserving essential multi-material characteristics of the target problem. Third, we train and deploy a residual RL policy, with discretized force observations and domain randomization, enabling robust zero-shot sim-to-real transfer and the first demonstration of a learned policy that debones a real chicken shoulder. Our experiments in our simulator, on our physical testbed, and on real chicken shoulders show that our learned policy reliably navigates the joint gap and reduces undesired bone/cartilage contact, resulting in up to a 4x improvement over existing open-loop cutting baselines in terms of success rate and bone avoidance. Our results also illustrate the necessity of force feedback for safe and effective multi-material cutting. The project website is at https://sites.google.com/view/chickendeboning-2026.
Problem
Research questions and friction points this paper is trying to address.
Automating precise chicken deboning while avoiding bone contact risks
Developing adaptive force-feedback cutting policy for multi-material joints
Creating simulation and physical testbed for safe robotic cutting evaluation
Innovation
Methods, ideas, or system contributions that make the work stand out.
Learning-based dynamically-adaptive 6-DoF cutting policy
Open-source simulator for multi-material cutting training
Residual RL policy with robust sim-to-real transfer
🔎 Similar Papers
2024-04-28arXiv.orgCitations: 15
