KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Sensor Fusion
Process of integrating data from multiple sensors to produce a more accurate and robust environmental estimate than would be possible with individual sensors alone. This technique combines the strengths of different sensor types to compensate for their respective weaknesses.
Lidar
Laser detection and ranging technology that measures distances by emitting laser pulses and analyzing reflected light to create accurate 3D representations of the environment. Lidar is essential for mapping, obstacle detection, and autonomous navigation.
Computer Vision
Field of AI enabling machines to interpret and understand visual information from digital images or videos. In robotics, it enables object recognition, motion detection, and analysis of complex scenes.
Visual Odometry
Technique for estimating a robot's displacement by analyzing successive changes in a sequence of images captured by its cameras. It allows calculating the robot's trajectory without using traditional motion sensors.
Point Clouds
Set of data points in a three-dimensional coordinate system representing the external surface of objects or environments. This data is essential for 3D reconstruction, obstacle detection, and trajectory planning.
Proprioceptive Sensors
Internal sensors measuring the robot's own state such as joint positions, wheel speeds, or chassis inclination. They provide crucial information about the robot's internal configuration and movement itself.
Exteroceptive Sensors
External sensors that collect information about the environment outside the robot, such as cameras, microphones, or distance sensors. They enable the robot to perceive and interact with its physical environment.
3D Mapping
Process of creating detailed three-dimensional representations of the physical environment from multiple sensor data. These maps enable robots to navigate, plan trajectories, and interact with their environment intelligently.
Obstacle detection
Perception system identifying objects and surfaces that could hinder the robot's movement in its environment. This critical functionality uses various sensors to ensure safe and autonomous navigation.
Pattern recognition
System's ability to automatically identify and classify specific patterns or objects in sensory data. In robotics, it enables object identification, signal reading, and adaptation to structured environments.
Autonomous navigation
Set of technologies allowing a robot to move independently in an environment without direct human intervention. It combines perception, planning, and control to achieve complex navigation objectives.
Visual localization
Technique determining a robot's position and orientation using only visual information compared to a pre-existing map. It enables precise localization even in environments without positioning infrastructure.
Convolutional neural networks
Deep learning architecture specifically designed to process grid-structured data like images, widely used in robotic vision. These networks excel at object detection and semantic classification for perception.
Scene reconstruction
Process of creating a complete and coherent 3D model of a scene from 2D images or multiple sensor data. This technique allows robots to understand and interact with complex and dynamic environments.
Sensor signal processing
Set of algorithmic techniques for filtering, interpreting, and extracting relevant information from raw robotic sensor data. This processing is fundamental for transforming physical signals into perceptions usable by AI.
Visual SLAM
SLAM variant primarily using cameras as sensors to build maps and locate the robot simultaneously. This approach offers economical and lightweight solutions for autonomous navigation in various environments.