ChicGrasp: Imitation-Learning based Customized Dual-Jaw Gripper Control for Delicate, Irregular Bio-products Manipulation

2025-05-13

AI Summary

Automated grasping and hanging of poultry carcasses in processing lines remains challenging due to their slippery, fragile, and highly variable anatomical morphology. Method: This paper proposes a hardware-software co-designed end-to-end solution: (i) a pneumatically actuated, independently driven dual-jaw gripper for morphological adaptability; and (ii) a conditional diffusion policy controller tailored for biological soft-bodied objects, trained exclusively on 50 multi-view RGB + proprioceptive teleoperation demonstrations to achieve closed-loop grasp-lift-hang execution. Contribution/Results: Evaluated on live chicken carcasses, the method achieves a 40.6% improvement in grasp success rate and completes each cycle in 38 seconds—significantly outperforming IBC and LSTM-GMM baselines (both exhibiting 100% failure). It introduces the first conditional diffusion–driven manipulation framework for soft-bodied targets and fully open-sources the CAD models, source code, and dataset.

Technology Category

• Intelligent Robots: Manipulation
• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems

Abstract

Automated poultry processing lines still rely on humans to lift slippery, easily bruised carcasses onto a shackle conveyor. Deformability, anatomical variance, and strict hygiene rules make conventional suction and scripted motions unreliable. We present ChicGrasp, an end--to--end hardware--software co-design for this task. An independently actuated dual-jaw pneumatic gripper clamps both chicken legs, while a conditional diffusion-policy controller, trained from only 50 multi--view teleoperation demonstrations (RGB + proprioception), plans 5 DoF end--effector motion, which includes jaw commands in one shot. On individually presented raw broiler carcasses, our system achieves a 40.6% grasp--and--lift success rate and completes the pick to shackle cycle in 38 s, whereas state--of--the--art implicit behaviour cloning (IBC) and LSTM-GMM baselines fail entirely. All CAD, code, and datasets will be open-source. ChicGrasp shows that imitation learning can bridge the gap between rigid hardware and variable bio--products, offering a reproducible benchmark and a public dataset for researchers in agricultural engineering and robot learning.

Problem

• Control delicate irregular bio-products with dual-jaw gripper
• Overcome deformability and anatomical variance in poultry handling
• Bridge rigid hardware and variable bio-products via imitation learning

Innovation

• Dual-jaw pneumatic gripper for delicate bio-products
• Conditional diffusion-policy controller for motion planning
• Imitation learning from 50 multi-view demonstrations