Preface
Lidar is no longer an unfamiliar sensor in the field of intelligent driving. It is regarded by most car companies and intelligent driving companies as "a necessary sensor to achieve high-level autonomous driving." Because its unique 3-dimensional perception information can make up for the blind spots of other sensors such as cameras in very special scenes - blind filling. There are two main reasons why lidar is popular:
- Car companies have an urgent need for high-end assisted driving;
- LiDAR prices have dropped;
At present, domestic models equipped with lidar include: Xpeng P5, Weilai ET7, Jihu Alpha S, Audi, BMW, Zhiji, Nezha, etc.
1 Principle of lidar
LiDAR consists of three parts: laser transmitter (Laser), receiver (Receiver), and processor (Processor) . Its ranging methods mainly include: time-of-flight method and space geometry method.
1.1 Time of Flight method TOF (Time Of Flight)
- dTOF (direct time of flight method)
When the laser emitter accurately irradiates the surface of any object, reflection will occur. A beam of light must return to the starting point and be received by the receiver . The processor calculates the round trip time of this beam of light and can obtain the precise distance of this point.
- iTOF (indirect time of flight method)
Ranging is measured by amplitude modulating the continuous laser and measuring the return time of the modulated light. Also called phase laser ranging. Within a certain distance, iTOF has higher accuracy and lower cost, but the output frequency is lower than dTOF. As the distance increases, the accuracy will become lower.
1.2 Space geometry method
The laser is emitted at a certain angle, and the receiver is in the shape of a strip. When the emitting angle and the distance between the emitting point and the receiving point are known, the distance of the measured object can be calculated. This ranging method has low cost and limited accuracy, and is widely used in the field of surveying and mapping.
2 Evaluation indicators of lidar
LiDAR is a sensor that can provide depth information. Its performance indicators are mainly reflected in: ranging capability, resolution and accuracy. The ranging capability is determined by the overall efficiency of the transceiver module (composed of laser transmitter, receiver and optical module); the resolution is determined by the signal processing capability of the transceiver module; and the accuracy is determined by the number of laser transmitters in the transceiver module.
Currently, in terms of resolution, NIO has launched a fixed gaze function on its model ET7: the bionic human eye flexibly zooms and can generate a high-density point cloud with an angular resolution of 0.06°*0.06° in key driving areas, making it possible to target targets in this area. See more clearly.
The accuracy of lidar is measured by POD (Probability Of Detection, Probability of Detection) .
POD = number of theoretical points/number of effective points
Theoretical number of points - generally the number of laser beams emitted for more than 100 consecutive frames;
Number of effective points-----the number of detected laser beams;
POD reflects the ability and stability of lidar to receive returned points and is an important indicator of lidar performance;
references: