Pick[+]
Pick[+] is BitMetrics’ AI-enabled solution for Pick Place and Random Bin Picking processes. Non-expert users in computer vision or artificial intelligence can autonomously create and modify their own applications, without requiring additional integration or coding. BitMetrics currently offers a cost-effective integration with Intel®️ RealSense’s camera. Integrations with Zivid©️ and IDS’s Ensenso©️ 3D cameras are also available for applications in high accuracy contexts (contact us for further details on these options). Pick[+] is a Deep Learning-based solution especially apt for chaotic, random positioning of items. It works both with: known parts using texture or geometric object attributes to perform a specific place action. It can also work with different pieces in the same pick scene with differentiated robot actions for part, unknown objects (parts that are automatically recognized without requiring CAD or any other user input). Once the scene is defined, Pick[+] is plug play offering candidate selection in milliseconds. Relevant advantages: It provides flexibility to your robot: Pick[+] license allows the users to create as many applications as needed (one application is executed at a time) Different pieces with different robot actions can co-exist in the same application It is easy to implement by following a guided set-up method for the picking process. All you have to do is define the robot’s type of ‘place’ that meets the specifics of your process Pick[+] is based on BitMetrics’ proprietary algorithms and models to ensure it performs in changing environments that require flexibility. Its license enables users to create as many applications as they need, which makes Pick[+] a competitive and reliable option to work with different parts and processes.
Overview
Pick[+] is BitMetrics’ AI-enabled solution for Pick Place and Random Bin Picking processes. Non-expert users in computer vision or artificial intelligence can autonomously create and modify their own applications, without requiring additional integration or coding. BitMetrics currently offers a cost-effective integration with Intel®️ RealSense’s camera. Integrations with Zivid©️ and IDS’s Ensenso©️ 3D cameras are also available for applications in high accuracy contexts (contact us for further details on these options). Pick[+] is a Deep Learning-based solution especially apt for chaotic, random positioning of items. It works both with: known parts using texture or geometric object attributes to perform a specific place action. It can also work with different pieces in the same pick scene with differentiated robot actions for part, unknown objects (parts that are automatically recognized without requiring CAD or any other user input). Once the scene is defined, Pick[+] is plug play offering candidate selection in milliseconds. Relevant advantages: It provides flexibility to your robot: Pick[+] license allows the users to create as many applications as needed (one application is executed at a time) Different pieces with different robot actions can co-exist in the same application It is easy to implement by following a guided set-up method for the picking process. All you have to do is define the robot’s type of ‘place’ that meets the specifics of your process Pick[+] is based on BitMetrics’ proprietary algorithms and models to ensure it performs in changing environments that require flexibility. Its license enables users to create as many applications as they need, which makes Pick[+] a competitive and reliable option to work with different parts and processes.
There are several documents and video tutorials that cover some of the most common functionalities of Pick[+] and serve as a complement to our user manuals.
Specifications
Software version required
Pick[+] URCap (release version 1.3.0) supports Polyscope 5.10 and higher
The URCAP implements all communication with the external PC supplied with Pick[+] in the form of program nodes. The external PC runs the current version of Pick[+] (v.1.3). These nodes make implementing Pick[+] into the robot program as easy as possible.
DimensionsBOX A Length 55 cm x Wide 35 cm x Heigh 30cm – 5kg
Box B Length 45 cm x Wide 35 cm x Heigh 15cm – 3kg
DEPENDENCIESUser’s PC/Laptop for client’s installation and visualization, robot and gripper.
CERTIFICATIONS & STANDARD CE, RoHs
What's in the box?
Box A
Box A contains the following elements:
- PC with Pick[+] server installed.
- Ethernet cable for connection between PC and robot.
*Note: cable for connectivity between client PC and server PC is not provided.
Box B
Box B contains the following elements:
- Camera Intel RealSense D435i mounted on LED case.
- Camera adapter for eye-in-hand mode (camera in robot).
- Camera adapter for hand-eye mode (external camera).
- Calibration plate.
- Scanning mat.
- LED cable.
- Camera-PC cable.
- USB with Pick[+] client, URcap and user guide.
* Note: all mounting screws are included in the Box B.
License type One-time purchase
Features
- AI-based: it learns from user data (CAD models or RGB images of the pieces) by means of deep architectures thus allowing the generated models to be precise and robust for the Pick & Place task at hand.
- No human labeling: the data labeling and generation necessary for AI learning is automatically performed by Pick[+].
- CAD-independent: Pick[+] can work with unseen objects by identifying a feasible picking point for any item in real time based on surface analysis.
- Multi-category pick: Pick[+] can perform instance segmentation of several object categories in the same picking process.
- Flexible: different types of pick in the same application (i.e., automatic assessment of the pick point, pick points selected by the user on the 2.5D or 3D model of the object, etc.)
- Multiple alignment strategies: alignment based on geometric model, on texture or automatically assessed in relation to the principal axis of the object.
- Fast: multiple candidate selection on the millisecond range (this ensures successful picks within constrained cycle-times)
- Human-centric: easy to use, the user can build its own application with no need for vision expertise.
- Hardware agnostic: any camera, any gripper, any robot.
Compatibility
- UR3e
- UR5e
- UR10e
- UR12e
- UR16e
- UR20
- UR30
Availability
- France
- Italy
- Portugal
- Spain