Key takeaway: A robot that cannot sense its surroundings cannot act safely or effectively, and perception — using sensors to understand the world — is the foundation of capable and safe robotics, and it is improving rapidly with better sensors and learning.
Why Perception Matters
A robot acts on the world, but to act it must first understand the world. It needs to know where it is, what is around it, where objects are, and what they are. This understanding comes from perception — using sensors to sense the environment and software to interpret what the sensors detect.
Without good perception, a robot cannot navigate safely, pick up objects reliably, or avoid obstacles. Perception is the foundation of capable and safe robotics, and it is one of the hardest problems in the field.
The Sensors Robots Use
| Sensor | What it detects |
|---|---|
| Cameras | Visual information about the world |
| LiDAR | Distance and shape using light |
| Radar | Distance and motion using radio waves |
| Touch sensors | Contact and force |
Robots use a range of sensors. Cameras provide visual information about the world. LiDAR measures distance and shape using light. Radar measures distance and motion using radio waves. And touch sensors detect contact and force. Each sensor has strengths and limits, and robots often combine them.
How Perception Is Improving
Perception is improving rapidly because of advances in the sensors and the software that interprets them. Better sensors provide more accurate information, and learning-based software is much better at understanding what the sensors detect — recognising objects, understanding scenes, and predicting what will happen next.
These advances are making robots able to understand and act in the real world more reliably, which is expanding what robots can do.
The Bottom Line
Understand that perception is how robots understand the world around them, since a robot that cannot sense its surroundings cannot act safely or effectively, and perception is the foundation of capable and safe robotics. Robots use cameras, LiDAR, radar, and touch sensors, often combined, and perception is improving rapidly with better sensors and learning-based software that recognises objects and understands scenes — which is expanding what robots can do in the real world.
