Robot perception is the process by which a robot interprets sensor data from cameras, lidar, depth sensors and inertial units to build an actionable understanding of its environment, objects and its own pose.
Semantic Classification
Content
- Robot perception converts raw sensor readings from cameras, lidar, depth sensors and inertial units into representations of objects, surfaces and the robot’s own pose. It supports tasks such as detection, recognition and localisation.
- Perception pipelines combine signal processing, geometry and machine learning. The resulting environment models feed planning and control, enabling navigation, manipulation and interaction.