See With Selectin Machine Vision Software From Energid Technologies
Selectin provides a simple and powerful programming toolkit enabling robots to see their workspace and environment.
Robotic systems today have cameras but can’t see. Obsessive to a fault, they need precise lighting and object placement. Introduce any small variation, and they’re prone to malfunction.
See. Although manufacturers don’t tell you this upfront, most machine vision software packages today also require exquisite controls over lighting and object placement. This translates to delicate and fussy robotics systems in real-world settings that produce errors, downtime, and unanticipated expenses and delays.
Energid’s Selectin machine vision system enables the robot and its controller to envision a new workspace and a new way of automating. For starters, Selectin operates in real, three-dimensional space. (Most machine vision solutions “see” only in 2D and get the flat-footed and flat-sighted results you’d expect.)
Because Energid’s Actin™ control and simulation systems are so adaptive, Energid has developed a suite of similarly responsive Selectin tools – enabling, for instance, smart guidance for autonomous grasping of robot hands integrated with real-time visual feedback.
Selectin is a suite of C++ tools for machine vision, ready to be rapidly developed and deployed for custom inspection and vision-guided robotics solutions. Selectin solutions can be delivered as
- Turnkey inspection systems with a custom user interface
- Visual servoing modules for pick-and-place assembly or inspection – or any other application requiring autonomous grasping
- A machine vision engine that runs under your user interface
- An extensible Selectin toolkit to modify for your custom robot vision capabilities.
3D Object Tracking.
- 3D recognition and tracking – Tracks position, orientation, and articulation of targets from CAD models or image-derived trained views
- Speed – Fast tracking is possible using today’s low-cost GPUs for signal processing
- Flexibility – With its XML-based scripting language, Selectin operates in varied lighting conditions and supports varied sensor inputs: color, gray-level, hyperspectral, ladar, time-of-flight, and almost any 3D sensor that produces dense depth data
- Code reuse – Control algorithms developed for one application can be used on others
Articulated Object Tracking and Body Tracking.
The Selectin toolkit offers unprecedented 3D object recognition and tracking. Developers can build their own object recognition systems, training the algorithms via CAD models, photographs, or computer-generated images. Armed with this knowledge, Selectin systems can operate in various lighting conditions, distinguish between objects, and adapt to the position, orientation, and articulation of objects in its scene.
Selectin’s C++ modular software system is extensible and flexible. It works across a broad range of sensors – including gray-level, color, hyperspectral, multi-camera, ladar and swept linescan. And it can optimize run-time vs. accuracy at every step in the algorithm.
Crop Detection for Robotic Harvesting.
Selectin 3D object recognition consists of a spatial processing unit and a temporal processing unit. The spatial processor determines the type and pose of the object on a per image basis, and the temporal processor uses a track centric multiple hypothesis tracker (MHT) to track objects over time.
The spatial processing unit can be further broken down into three stages: The segmentation processor, the Initial Type, Pose, Geometry (ITPG) estimator, and the Refined Type Pose Geometry (RTPG) estimator. This ITPG returns several initial estimates of the target state using a subspace (Hyperspherical Harmonic) method and the RTPG refines these estimates using fast GPU based correlations and a non-linear optimization algorithm.