Advancements in Poultry Processing Robotics: Introducing ChicGrasp

The demand for innovative solutions in poultry processing has surged due to labor shortages exacerbated by the COVID-19 pandemic. In response, researchers at the Arkansas Agricultural Experiment Station have unveiled ChicGrasp, a groundbreaking robotic system designed to handle chickens with unprecedented efficiency.

Revolutionizing Poultry Processing

ChicGrasp employs a dual-jaw gripper with advanced imitation learning algorithms and camera perception capabilities. This enables the robotic system to mimic human movements, allowing it to securely grasp chicken carcasses by the legs and transfer them onto shackle conveyors for further processing.

“Embodied AI empowers us to create intelligent robots capable of interacting with real-world environments,” explains Dongyi Wang, the project leader and assistant professor in the Departments of Biological and Agricultural Engineering and Food Science. “We aim to apply similar principles used in autonomous vehicles to enhance chicken processing operations.”

Funding and Support

This innovative project has received substantial backing, including a $1 million grant from a collaboration between the U.S. Department of Agriculture’s National Institute of Food and Agriculture and the National Science Foundation. The Arkansas Agricultural Experiment Station is the research entity of the U of A System Division of Agriculture.

The findings detailing the development of ChicGrasp have been published in the journal Advanced Robotics Research, with open-source access to all computer-aided design files, code, and datasets, which promotes reproducibility in agricultural robotics research.

Innovative Learning Techniques

Traditional robotic methods, often reliant on programmed motions or suction cups, struggle with the unpredictability of a poultry processing line where birds vary in size, temperature, and posture. Wang’s team has pioneered an approach wherein the robotic system learns from human instructors, allowing for more effective and adaptive operations.

Graduate student Amirreza Davar played a pivotal role in designing the gripper and tailoring imitation learning to suit the robotic framework. “The human instructor provides a trajectory, giving the robot a solid starting point for learning,” Davar explains, noting this method significantly enhances accuracy and efficiency.

Continuous Learning and Improvement

The innovative use of the diffusion policy algorithm, introduced in 2023, allows the system to continuously refine its grasping strategies. This adaptive framework formulates robot control as a “conditional denoising process,” making the robotics capable of handling the unpredictable nature of poultry processing.

“Unlike traditional robots that are programmed for specific tasks, our approach enables the robot to adapt its actions based on real-time scenarios,” says Davar.

Addressing Speed Challenges

Currently, ChicGrasp showcases an impressive 81% success rate; however, speed remains a critical challenge. While a human can efficiently transfer a chicken carcass in approximately three seconds, the robotic system takes about 38 seconds to complete the same task.

To close this speed gap, researchers plan to implement enhancements to both the robotic arm’s velocity and the operational algorithms to minimize idle times.

Open-Source Innovation

The team’s decision to share hardware designs and training data aims to accelerate advancements in agricultural engineering, an area historically lacking reproducible datasets and benchmarks.

The study, “ChicGrasp: Imitation-Learning-Based Customized Dual-Jaw Gripper Control for Manipulation of Delicate, Irregular Bio-Products,” highlights collaboration among multiple graduate students and faculty from the University of Arkansas and Purdue University, with Wang as the corresponding author.

Conclusion

ChicGrasp represents a significant leap forward in robotic assistance for poultry processing, combining advanced learning techniques with practical applications in an ever-evolving industry. The ultimate goal remains to enhance efficiency and adaptability in agricultural operations while responding to the challenges of the modern workforce.

For more information on agricultural and food research in Arkansas, visit the Arkansas Agricultural Experiment Station. Follow the station on LinkedIn and subscribe to their monthly newsletter, the Arkansas Agricultural Research Report.

About the Division of Agriculture: The University of Arkansas System Division of Agriculture aims to enhance agriculture, communities, and families through research-driven best practices. With offices in all 75 counties and faculty across three campuses, the Division of Agriculture continues to lead in agricultural innovation.

This work has been made possible through various awards from the U.S. Department of Agriculture’s National Institute of Food and Agriculture, in collaboration with the National Science Foundation.

Source: The University of Arkansas