LiDAR – greater safety, efficiency and autonomy in the vehicle
Innovative LiDAR sensors are a key technology in the automotive sector of the future. Countless automotive processes which play a role in autonomous driving would be difficult to implement without the support of LiDAR. Thousands of individual distance measurements are composed of 10 to 30 3D images per second, and generate an exact three-dimensional image of the vehicle’s surroundings. Similarly to radar readings, the sensors help detect obstacles and measure distances.
What is LiDAR – Light Detection and Ranging
In contrast to radar measurements, as the term “light” in its name suggests, LiDAR uses an optical medium to take measurements.
This involves laser beams which are emitted by a sensor and reflected by an object. The laser beams are then received back by the sensor and analyzed by a connected processor.
Different technological methods for analyzing laser beam emissions are available.
How does LiDAR technology work?
The currently most widely used method is known as Time of Flight (ToF).
This is where a pulse of infrared laser light (850 and 905 nm) is emitted, and the length of time is measured until the reflected beam is detected again by the sensor. The challenge here is that the infrared parts of the same wavelength that are present in daylight have a negative impact on the beam due to noise. For this reason, the laser pulses are given a code, similar to Morse code. Only received signals that have the same code are analyzed.
Irrespective of this, a very high light output is required for the laser beam so that the emitted pulse is not swamped with noise at longer distances. However, the transmitting power must be limited in order to protect people’s eyes. As a result, the range of the ToF method is limited. The aim is to also be able to detect black cars from a distance of 200 meters. Currently, the ranges are at a maximum 80 to 100 meters.
In order to determine the speed of moving persons or objects, this method requires several measurements to be taken. Currently, algorithms (firefly process) are being worked on to increase the range of the ToF method.
The firefly process is supported by hypotheses tests which determine which pixels are realistic and which are not in noisy signals using pixel movement patterns. This means that only “real” measurement points behave according to the limits of physics. “False” measurement points can then be filtered out.
A more recent process is the Frequency Modulated Continuous Wave (FMCW) method.
FMCW is already known from radar technology, but is not yet available in series production for LiDAR.
With the FMCW method, the emitted laser beam (1550 nm) is constantly modulated and “chirped”. This means that the frequency of the signal is periodically raised and lowered. If the laser beam meets an object, it will be reflected on it. Due to time and the periodic change in frequency, the frequency of the received beam differs from the beam that has just been emitted. The difference between these two frequencies is proportional to the distance, and therefore provides information about the distance to the object. If the object moves, an additional frequency shift occurs due to the Doppler effect. Thus, the speed of an object can also be determined.
Systems with a wavelength of 1550 nm (FMCW-LiDAR) are more resistant to solar radiation and can work with a significantly lower light output which is safer for the eyes (40 times better eye safety). FMCW-LiDAR also offers a higher resolution than ToF-LiDAR.
The first LiDAR sensors were extremely expensive (> €10,000) and mechanically elaborate. This meant that serial application in vehicles was highly doubtful. Rotating mirrors directed the laser beam in the desired direction, line by line – like with a scanner. The mechanism required for this was very large and heavy, and required the greatest precision for the rotating parts.
These scanners were also very sensitive to vibration and impact. Only after solid state sensors were developed which operate without requiring any mechanical parts did the pathway to series vehicle application appear to have been cleared. In the future, the price should settle within the 3-digit range.
Transmitters and receivers are used for about 80 lines and with more than 100 points per line on sensor fields, some of which are the size of credit cards. 25 images per second can be captured in this way.
If several such sensors are strategically positioned around the vehicle, this enables a 360-degree, all-round view.
Advantages and disadvantages of LiDAR sensors
LiDAR sensors are able to generate a complete three-dimensional image of a vehicle’s surrounding environment.
This image leaves no room for interpretation and produces real 3D data.
With LiDAR, the distance and speed of objects can be determined very precisely.
However, LiDAR sensors have two major disadvantages:
- They cannot detect color which makes reading traffic signs impossible.
- They have a problem in fog – the damp air prevents the infrared laser from delivering reliable data.
LiDAR sensors are an essential component in making autonomous driving possible, and safer.
However, in practice, this technology also has many disadvantages. Most vehicle manufacturers agree that a combination of LiDAR, radar, camera technology and GPS (sensor fusion) is required to be able to autonomously operate cars in the future, even in difficult traffic situations.