NSF CAREER Award to Study Use of Robotics to Handle Soft Foods

Autonomous robots may be able to drive through rush hour traffic in L.A., which can test the sanity of the even the most skilled human drivers, but robots have yet to master the science of grasping soft foods like meat products or fruits with the speed, delicacy and precision of human hands.  

Differences in shape, texture and material properties of agrifood products limit the effectiveness of conventional automation approaches that are successful in other manufacturing environments. But periodic labor shortages and challenging working conditions for employees suggest chicken and soft food production might benefit from improved autonomous manufacturing systems. 

Enter Donyi Wang, an assistant professor of agricultural and biological engineering with both the U of A and the U of A System Division of Agriculture. The National Science Foundation awarded Wang a $511,074 CAREER award to develop autonomous manufacturing systems to handle soft foods with adaptability and precision. 

"By enabling robots to perceive and respond to product texture without physical contact, the work supports autonomous food processing, improved product quality and resilient domestic food supply chains," Wang wrote in his proposal. "The project contributes to the national interest by strengthening manufacturing competitiveness and advancing the scientific foundations of automation in a sector essential to the public." 

Wang will focus on both software and hardware development, primarily through customization of an off-the-shelf robotic arm. He will utilize hyperspectral imaging sensors to categorize chicken and berries. This noninvasive technology entails shining a special hyperspectral light on an object that reveals chemical signatures. In the case of produce, reflected electromagnetic wavelengths can be used to indicate protein, moisture and fat values.  

Take enough readings and run them through a texture analyzer, and eventually, you can determine the texture and softness of an object by the wavelengths it emits without having to even know exact chemical information. The relative softness of a product can be determined by texture alone, which in turn will tell the robot how much force to apply when gripping it. 

To teach the robots how to manipulate soft foods, Wang will use an advanced imitation learning algorithm and camera perceptions, in which humans demonstrate how to handle objects through computer sticks or paddles, providing a ground truth for the robots to build on. He has been developing this technology for several years, using it to invent the ChicGrasp, a "dual-jaw robotic gripper with pinchers that can grasp a chicken carcass by the legs, lift and hang it on a shackle conveyor to be moved on for further processing." Introducing hyperspectral sensing into the control pipeline will improve the system's reliability and adaptability. 

The third and last component of Wang's award will be ensuring the system can take what it learns from grasping chickens and berries and apply it to other commodities it has never seen without extensive retraining. Oranges, apples and pears will be ripe for the grasping.  

CAREER awards are the NSF's most prestigious award for early career faculty who have the potential to serve as academic role models in research and education and to lead advances in their department or organization. The awards are for five years and include teaching and public-outreach components. This award will help cement the foundation of Wang's career. 

Wang holds a joint appointment in the College of Engineering and the Department of Food Science for the Dale Bumpers College of Agricultural, Food and Life Sciences. He also conducts research through the Arkansas Agricultural Experiment Station, which is the research arm of the U of A System Division of Agriculture.