What are the latest advances in radar and LiDAR technologies for sensor fusion: part 2

In contrast to the widespread use of radar sensing, automotive applications for light detection and ranging (LiDAR) sensors are much more limited. This is due in part to their bulky size (especially electromechanical LiDAR) and high cost. Similar to radar designs, except that LiDAR uses lasers instead of radio waves for detection and ranging, it is a crucial technology for pedestrian detection, monitoring complex overtaking maneuvers, and detecting obstacles ahead of the vehicle.

Overview of new LiDAR technologies

Electromechanical LiDAR that has 64-channels based on a rotating mechanical design has been the dominant LiDAR technology for driverless vehicles for many years (see Figure 1c). They offer a 360° view horizontally, with varying ranges vertically, and a scanning speed of one million samples per second at 15 Hz. This established technology is being challenged by solid-state LiDAR designs. Solid-state LiDAR designs use microelectromechanical systems (MEMS), optical phased arrays (OPAs), and even Flash LiDAR, or fast light detection and ranging.

In MEMS solid-state LiDAR, instead of moving the laser position mechanically across the ﬁeld of view (FoV), it uses a stationary laser and integrated electromechanical mirrors whose tilt angle can be adjusted by an applied voltage (Figure 1d).

OPA systems use phase modulators to direct laser beams, similar to the operation of phased array radar (PAR). In an OPA system, the optical phase modulator controls the speed of light passing through the lens. The technique enables control of the optical wave-front shape, as shown in Figure 1b. The bottom beam is not delayed, but the middle and top beams are delayed by increasing amounts. This type of design effectively “steers” the laser beam to point in different directions.

Fast light detection and ranging, or Flash LiDAR systems, work similarly to digital cameras by using a burst of light to illuminate the environment and photodetectors to collect the backscattered light (Figure 1a). It allows faster data capture and is highly resistant to light distortion.

The impact of LiDAR advances on sensor fusion performance

Several companies offer solid-state LiDAR systems and components based on these different technologies. Researchers are investigating different ways of improving designs to increase performance and reduce costs.

One mass-producible Chinese manufacturer’s digital LiDAR system exceeds 500 beams and supports customization from 520 to 2160 beams. In addition to securing main LiDAR design-ins from several leading North American Robotaxi customers, customized versions are already in mass production on some popular Chinese vehicles.

From the research perspective, one group of researchers has demonstrated a MEMS mirror-based 360° LiDAR system with high angular resolution. As shown in Figure 2, their design combined a MEMS mirror with a rotation platform for the LiDAR system to achieve a 360° × 8.6° (horizontal × vertical) FoV. Compared with existing commercial multi-channel 360° LiDAR systems (shown in Figure 2), their system demonstrated 13.8 times better angular resolution than a 64-channel electromechanical LiDAR sensor and delivered an angular resolution of 0.07° × 0.027° (horizontal × vertical).

Part 3 of this blog will discuss additional design techniques that are also implemented or being pursued for improving sensor fusion in future automated vehicles. Read part 1 here.