CREATIVE ALGORITHMS AND SENSOR EVOLUTION LABORATORY
Our AI and knowledge driven semantic solution makes your robot possible humanlike high-level navigation and manipulation for more efficient manufacturing and better service.
Semantic Analyzer
Analyze object data and produce its semantic descriptor.
Use cases : model learning, object recognition, etc.
Semantic Localizer
Localization of robot based on semantic understanding of environment.
Use cases : location based service, place recognition, path finding
Semantic Knowledge Description Language(SKDL)
Interpret and inject domain knowledge into semantic model of workspace.
Use cases : verbal or script expression of semantic-episodic domain knowledge for factory, home, office, and outdoors.
Semantic Modeler(SM)
Build the domain specific knowledge based semantic model of environment data from sensors including object, place, occupant, and robot.
Use cases : domain knowledge injected description of environment via image or point cloud data conversion.
Semantic Recognizer
Recognition of environment using semantic descriptor and deep learning.
Use cases : object recognition, place recognition, face recognition, scene understanding, etc.
Semantic Map Builder
Build 2D/3D semantic map by combining semantic model data and scanned odometer data
Use cases : semantic topology map, semantic metric map, semantic object map, etc.
Navigation with semantic map and robot sensor data. It includes the core functions of semantic map builder, semantic localizer and recognizer.
Use cases : factory automation robot, logistic robot, home-office service robot, field robot, aerial robot, underwater robot, etc
Manipulation of object based on semantic modeling, analysis, and recognition
Use cases : automation of manufacturing tasks, logistics, delivery robot, product handling robot
Semantic visual servo for tracking of object based on learned semantic model and recognition
Use cases : automation of object tracking for personal/public/field robot service and manufacturing.
Mission planning and execution based on behavior modeling and control of robot under semantic modeling and recognition framework
Use cases : mobile manipulator control, multiple robot control, leader-follower robots, mission execution for personal/public/field frobot service and manufacturing
sensor model-segmentation-classification-detection
-sensor dependent scene grouping (ex. scene graph)-execution map booking-semantic descriptor gen-deep learning
Use cases: semantic DB, semantic map, execution map, perception, place recognition, scene understanding.
Item | Existing approaches | CPKit solution |
---|---|---|
Level of intelligence | Sensor data driven | Semantic knowledge driven |
Grid map | Occupancy rectangle grid | Semantic triangle –hexagonal grid |
Modeling | Appearance model | Semantic ontological model |
Sensor characteristic | Sensor independent | Sensor dependent |
Localization | Robot Location | Robot Location & Object-place recognition |