Roboception Introduces ItemPickAI Allowing Picking and Oriented Placing of ‘Unseen’ Objects

Roboception has introduced ItemPickAI, an advanced AI-driven software designed to enhance robotic pick-and-place applications. This cutting-edge solution enables automated grasping and precise placement of ‘unseen’ objects—items that the system has not explicitly learned through pre-trained models or prior data.

With its powerful AI-based grasp computation, ItemPickAI enables the efficient separation and oriented placement of unknown and deformable objects within a specific category. Applications such as palletizing, depalletizing, and unpacking of bags, sacks, and pouch packaging can be performed with exceptional accuracy and efficiency.

AI-Driven Grasping and Placement

ItemPickAI calculates optimal gripping poses for suction grippers, even when objects are randomly arranged in chaotic bins, load carriers, or mixed and unmixed configurations. The software ensures that robotic systems can handle soft, deformable items with varying shapes, sizes, and fill levels.

At the core of this system is a vision module consisting of a 3D stereo sensor and Roboception’s rc_cube computing platform. This module enables precise object detection, segmentation, and grasp pose calculation. Thanks to neural network-based segmentation, ItemPickAI can reliably detect objects with uneven surfaces and deformable contents, ensuring that the robot selects an ideal grasping point every time.

Seamless Integration and Adaptability

One of the standout features of ItemPickAI is its easy configuration. During commissioning, users only need to select the object category and specify the size of the gripper’s suction cup. These parameters can be flexibly adapted to new conditions at any time, allowing for fast implementation of pick-and-place operations – even without expert knowledge of AI or image processing.

ItemPickAI runs offboard with any 3D sensor via Roboception’s rc_cube and is easily configured using a standard interface. The software computes a configurable number of grasp poses for all detected objects within a predefined workspace.

Enhanced Object Handling and Workflow Efficiency

ItemPickAI ensures robust object segmentation, even for items with irregular surfaces and varying contents, using AI-powered neural networks. This guarantees that grasp points are optimally placed at the center of each segmented object, while also determining object orientation to enable precise placement.

The system is highly versatile and can be used with any standard suction gripper, as the size of the suction device can be customized to match specific application needs. The working area can be defined either automatically—by recognizing a container—or manually, by specifying the area of interest.

For seamless integration with robotic systems, ItemPickAI comes with sample programs that simplify its deployment into robot controllers.

Optimized for Logistics and Industrial Automation

Beyond traditional automation, ItemPickAI is particularly well-suited for logistics applications. Robots equipped with this software can efficiently pick, align, and place soft or disordered objects from crates, streamline depalletizing operations, and transfer items onto conveyor belts.

For optimal performance in pick-and-place tasks, the rc_randomdot projector is recommended, especially when load carriers need to be accurately detected during the pick process.

AI-Powered Adaptability for the Future of Robotics

With ItemPickAI, Roboception brings AI-driven flexibility and intelligence to robotic pick-and-place applications. By enabling fast, adaptable object handling, the software paves the way for smarter, more versatile automation solutions across industries.

For more information: www.roboception.com