The Lasers Used in Self-Driving Cars

Interview conducted by Vignesh MukundJul 30 2018

Image Credits Shutterstock/temp-64GTX

The self-driving car has several sensors that detect obstacles and map its environment. Amongst these sensors, the LiDAR is most like the human eye in its function. It’s a detection and ranging system much like the radar but instead of radio waves, the LiDAR operates on light waves.

LiDAR systems illuminate the target area with a pulsed laser signal and calculate the time it takes for the reflected signal to be returned to the receiver. These systems comprise of a laser source, a photodetector, data processing electronics, and motion-control equipment.

There are broadly two types of LiDARs used in autonomous vehicles. These are classified based on how they scan their surroundings. The 3D flash LiDAR consists of a wide-angle source and wide-angle optics to focus all the reflected light from one exposure onto the detector. This information is then used to model the surroundings based on time-of-flight calculations. The scanning LiDAR is a system that emits light sequentially in every direction and detects the echoes one by one to map its surroundings.

The metrics that are used to gauge the performance of a LiDAR system are:

• lateral resolution
• longitudinal resolution
• long-range detection
• scan rate

Of the different components in a LiDAR, the laser source is most critical for performance. The quality of the laser beam and its divergence is crucial for high lateral resolution. A short laser pulse duration and low timing jitter ensure good longitudinal accuracy. Pulse energy is key to attaining long-range detection, while high pulse repetition rates allow faster scanning leading to a higher data throughput.

Wavelengths between 600 nm and 1000 nm are usually used for non-scientific applications. The human eye is sensitive to these wavelengths and can easily absorb them. The power at which the laser has to operate to map obstacles at long ranges is beyond what is considered eye-safe. LiDAR systems that operate at 905 nm are, hence, limited in their range. Infrared lasers of 1.5 um are a great alternative for detecting solid bodies and obstacles at long ranges, and since they are not absorbed by the human eye, they are suitable for high-performance applications. In addition, the atmosphere is largely transparent at this wavelength, and highly efficient detectors are available.

Another aspect of the laser source that's critical for good performance is the pulse duration. A short pulse of the order of a few picoseconds is highly desirable to achieve longitudinal resolutions of a few millimeters to centimeters. However, such a short pulse would cause the broadening of the laser spectrum, thereby decreasing the signal-to-noise ratio (SNR). While pulses longer than 1 nanosecond eliminate the noise, they also lower the resolution of the system. To balance high-longitudinal accuracy with good SNR, current systems have pulse durations of the order of a few hundred picoseconds.

Laser-diodes and fiber-lasers are both commonly used sources in LiDAR systems. Laser-diode sources are usually vertically stacked, meaning the incoherent addition between the layers in the stack often leads to laser powers exceeding that of a class 1 eye-safe laser. By replacing the vertical stacks with a vertical-cavity surface-emitting laser, the total output power can be decreased, but this consequently lowers the maximum range at which these systems can function. Fiber-lasers, on the other hand, offer high pulse-repetition rates of 5 kHz at power levels of 10 W to 250 kHz at power levels of 300 W. However, these are expensive systems compared to pulsed-diode systems.

The ideal LiDAR system must contain a laser source at an eye-safe wavelength with sufficient power to detect dark obstacles at a distance of 100 m with 10 cm accuracy, and operate at temperatures between -40 C to 85 C.

Sources:

1. Lasers for Lidar: Application parameters dictate laser source selection in lidar systems. https://www.laserfocusworld.com/articles/print/volume-53/issue-03/features/lasers-for-lidar-application-parameters-dictate-laser-source-selection-in-lidar-systems.html