Processing Produce With Artificial Intelligence
Combining custom hardware with vision and AI
Motivo’s client is a wholesale distributor of processed fruit and vegetables located in the USA. Alignment is critical when slicing or coring the client’s products; an incorrect slice can lead to tossing an entire fruit away and greatly influence the viability of the business. This was a task that only humans could do before as it required human dexterity to rotate the produce and human knowledge to recognize the correct orientation for fruits and vegetables of varying sizes. The client needed a system to automatically sense and orient different produce varieties and shapes to ensure correct alignment before slicing product at a pace that matched or surpassed current manual labor speeds.
Motivo’s solution needed to accommodate different types of produce with a wide variety of shapes, sizes, and imperfections. It needed to be robust and food safe with high cycle times and intense usage. Performance was also key; high precision and accuracy was needed to reduce the need for post-process labor.
OEM automated solutions existed but were limited to “perfectly shaped” fruits and vegetables and yet still yielded about 30% incorrect alignment. These solutions required downstream labor for rework. This labor-intensive rework is highly repetitive and hazardous due to close proximity to the industrial cutters.
Motivo began the project by looking at the existing process as-installed, and evaluating different opportunities up front and in coordination with the client. Motivo then proposed a solution that directly integrated with most existing industrial food slicers and cutters to minimize cost and space requirements.
Motivo’s sensing system uses artificial intelligence to visually determine the yaw orientation of a fruit or vegetable as well as its geometric center height, size, and other key parameters. A manipulation stage then gently grasps, lifts, and rotates the fruit until correct alignment is achieved.
The final solution was able to replace four humans working with/around razor sharp edges in an extremely loud environment (double hearing protection required). Not only was the solution successfully built and tested, but additional automation opportunities up and downstream were identified and are in the process of being implemented. This effort was intentionally designed to be modular in support of future expansion without impacting the flow of the production facility. This project went from initial ideation to a working proof of concept prototype with custom machine vision software in six months.