How AI Paves a New Path for Vision-Guided Pick and Place Robots 

Robots are Everywhere

From the grocery store to the farm to the factory floor — robots are everywhere. More than half a million robots shipped in 2023 and, while it’s not more than the record number hit in 2022, it’s close; more significantly, it highlights the growing importance of robots across a wide range of industries, far beyond just automotive manufacturing. Businesses worldwide are deploying robots not only to solve persisting labor shortages, but to increase productivity, boost efficiency, and drive revenue.

As industrial processes evolve to rely more on robots, these systems constantly face new challenges. In robotic pick and place environments, for example, robots are expected to handle greater variety and more exceptions in high-speed processes. Introducing AI-enabled software into vision-guided systems can deliver the flexibility required for those sorts of complex settings.

Exception Handling Help

Vision-Guided Pick and Place Robotics in Manufacturing and Logistics

Vision-guided pick and place robotic systems are a common type of robotic solution deployed in such settings as general manufacturing, logistics and warehousing, and food and beverage. For many companies in these industries, not only do labor shortages persist, but many of these jobs have been traditionally hard to fill. Pick and place systems typically consist of a 3D machine vision system, robotic manipulator, and software that allows the robot to identify, pick, and place various objects. Common examples include packaging, sortation, singulation, and exception handling, which involves dealing with unforeseen occurrences that can present challenges. 

In applications where the product remains the same size, shape, or color — such as plastic or metal automotive parts or consumer packaged goods — a pick and place system may not require AI capabilities. However, in certain applications — such as those dealing with high variability or organic products — or when attempting to handle exceptions and anomalies, AI adds the adaptability needed to solve these challenges. 

Tackling Variability and Anomalies with AI-Enabled Robotics

Combining AI-enabled inspection capabilities with robotic picking allows the system to reliably handle exceptions. In food and beverage environments, for example, exceptions might involve a packaged drink with a missing straw or a label facing downward, or a corn dog with a broken stick or an exposed hot dog underneath. AI-enabled software can identify anomalies in real time and determine an appropriate course of action, such as placing a good object onto another conveyor or moving an object with a defect to an exit.

Flexible Food and Beverage Processing

Vision-guided pick and place can also be difficult in food and beverage because of the high variability and organic nature of the products. Take poultry processing, for example. Traditionally, installing a vision-guided pick and place system that can singulate individual pieces, recognize size and orientation, and pick individual items from bulk involved installing large, expensive, and difficult to maintain singulation equipment.

AI and 3D Vision: Optimizing Poultry Processing with Precision

Integrating AI, 3D vision, and robotics, however, allows food processing companies to pick individual items from bulk while also inspecting for key attributes, allowing the robot to pick, orient, and place the item on a tray in an optimal way (e.g., placing a piece of chicken membrane side up). In doing so, companies improve their processes both upstream and downstream while cutting costs, saving floorspace, and increasing throughput.

Robotic Handling of Produce: Overcoming the Challenges with AI

Robotic handling of produce can also be tricky, but again, AI can help here as well. When a leading frozen foods supplier in the US was having quality challenges, they turned to industrial automation with 3D AI solutions to increase efficiencies and reduce waste.

Don’t Worry About Data Generation

Whether its food and beverage, consumer packaged goods, agriculture, or general manufacturing, companies looking to vision-guided pick and place systems that leverage AI do not necessarily need to worry about producing a vast image set for model training if that option does not exist or is not feasible. Oxipital AI offers a proprietary synthetic data generation pipeline that produces pre-trained object models without placing the onus on the customer for producing and labeling images. These models bring a deep object understanding to the system that offers the system performance and flexibility needed for today’s applications. 

AI isn’t going to solve all application problems, but for vision-guided pick and place, AI technology helps open new doors. Instead of wondering why or when you should adopt AI into your existing or new automation systems, contact Oxipital AI today to request a demo and see with your own eyes the power of AI-enabled pick and place technology.