Guiding the light

Share this on social media:

A small-sized silicon photonics chip that can be used for non-mechanical beam steering and scanning. (Credit: Yokohama National University).

Japanese researchers have employed a bespoke prism lens to reduce the size and cost of lidar technology.

Light detection and ranging (lidar) has been recognised as an effective means of granting 3D sensing capabilities to mobile systems such as autonomous vehicles, robots and drones for applications including transportation, logistics, security, mapping and surveying.

Many lidar systems comprise a laser source, a photodetector and an optical beam steering device. This device sweeps the laser’s pulsed beams across the surrounding environment, enabling 3D data points – and thus a 3D map of objects in the vicinity – to be generated by timing how long the pulses take to return to the photodetector. 

While many builds of these beam steering devices comprise a mechanical system such as a rotating mirror, these have been known to make the system large, costly and unstable1, which impedes their ability to be integrated into mobile systems on a large scale. 

Micro-electromechanical system mirrors have previously been explored as a means of size and cost reduction2 for mechanical beam steering, however these can result in a trade-off between size, beam divergence (or resolution), and speed. Therefore, complete non-mechanical (solid-state) lidar solutions are sought after, for example those under development by firms Innoviz, and Luminar, which have already attracted the attention of BMW and Volvo respectively, who now plan to integrate them into future models.

The past few years have seen heavy investment in firms such as Innoviz and Luminar, with more than $1bn in corporate and private investment being made across 50 lidar start-ups between 2017 and 2019. This includes $420m in 2018 alone, according to a Reuters analysis of publicly available investment data. It has allowed the technology to move forward significantly, hence it is now beginning to reach the required performance, size and price-point that has pricked up the ears of high-flyers in the automotive sector. 

Prism-powered solid-state lidar

It’s not just commercial start-ups that are working on developing solid-state lidar technology, as was demonstrated earlier this year by researchers from Yokohama National University in Japan.

They explained in a recent Optica paper that one approach to solid-state lidar engineers have turned towards in recent years is the use of optical phased arrays (OPAs), which are able to direct an optical beam without the use of mechanical parts. However, according to the researchers, such an approach can become complicated due to the sheer number of optical antennae required, as well as the time and precision needed to calibrate each piece. In addition, there are still many challenges for OPAs, including a trade-off between steering range, resolution and efficiency, as well as a need for complicated and power-consuming optical phase control.

Figure 1: (a) Top view of fabricated chip. (b) SEM image of LSPCW. Magnified view shows the third-row lattice shifts and shallow grating. (c) Prism lens loaded above the device. (Credit: Yokohama National University).

The researchers have previously explored an alternative approach to solid-state lidar by developing a beam steering device using a silicon lattice-shifted photonic crystal waveguide (LSPCW). As light interacts with the crystal, it slows down and is emitted to the free space in the form of a beam. The beam has a wide divergence (leading it to be called a ‘fan beam’ by the researchers), however a cylindrical collimator lens was used to convert it to a spot beam that could then be steered in the desired direction. 

This previous approach had some crucial problems, however, according to the researchers. It suffered from downward light emission loss and a large beam divergence, which led to severe internal reflection loss and collimation loss at the lens. In addition, the collimation condition was not maintained by the cylindrical lens for different beam angles, so manual adjustment of the lens position was necessary at each beam angle, in order to direct the beam correctly.

To overcome these issues while achieving wide-range two-dimensional beam steering, the researchers designed a shallow-etched diffraction grating and a bespoke prism lens, which has an aspherical design featuring a convex lens shape formed on two surfaces of the prism. The shallow-etched grating reduces the downward emission loss and reduces the internal reflection loss and collimation loss, while the prism lens collimates the fan beam and maintains the collimation condition for the targeted beam steering range. 

With the new setup, the researchers were able to steer a sharp spot beam with an average beam divergence of 0.15° in the range of 40° × 4.4° without requiring precise adjustment of the lens position. The number of resolution points obtainable with the setup was 4,256. 

The researchers were able to achieve conventional mechanical beam scans using their new device, such as a raster scan that would normally be conducted using a polygon mirror, or a zig-zag scan using a galvano scanner. In addition, because the system is completely non-mechanical and not affected by inertia from movement, the scanning speed does not fluctuate between the centre and the edge of the scan. What’s more, flexible scans, such as spirals and figures of eight, which are difficult to obtain by conventional mechanical systems, are also possible with the new system. This will extend its application fields beyond lidar, according to the researchers, for example to optical beam tracking, security systems, free-space communications, and more.

To fabricate the device, the researchers used a 200mm diameter silicon-on-insulator and a silicon photonics CMOS process that achieved a minimum feature size of <130nm by employing excimer laser exposure and a phase-shift mask. Image (a), below left, shows the fabricated device chip, which was 5.5 × 4mm2. At the centre, 32 LSPCWs of 1.2mm length are integrated with an 80µm pitch. Image (b) shows a scanning electron microscope image of the LSPCW. Its magnified view shows the formation of the shallow grating and uniform holes perforated after the shallow etching. Then (c) shows the prism lens, which is 24mm wide and 18.7mm high, and was formed using acrylic cutting.

Room for improvement

Next, the researchers plan to more fully demonstrate the potential of their device for solid-state lidar, as well as work on improving its performance with the ultimate goal of commercialisation. 

For example, regarding the prism lens, the researchers explain that using a smaller one would help enhance the steering range, however, aberration would increase at large angles. Therefore, the researchers believe a more sophisticated aspherical design and precise fabrication of the lens will be required. They noted in their paper that while a larger lens is space consuming and may be considered a disadvantage compared with the OPA approach to non-mechanical beam steering, this could actually be an advantage when applied to lidar: ‘In lidar systems, a large aperture that receives the returned light is crucial for high sensitivity,’ they explain. ‘It is only done by enlarging the chip size in OPAs, but is done by enlarging the [prism] lens in our device, which results in a high cost and cost savings, respectively.’ 

In addition to improving the design and fabrication of the prism lens, the researchers deduced that by using a longer waveguide and a shallower and more uniform grating, in addition to increasing the number of LSPCWs used from 32 to 256 in the same footprint – by integrating them as closely as possible – their device has the potential to achieve 345,600 resolution points.

The research was funded by the Japan Science and Technology Agency through an Accelerated Innovation Research Initiative.


CeNing Optics Featured product

CeNing, established in 2004, manufactures a wide range of precision prisms including right angle prisms, wedge prisms, dove prisms, penta prisms, corner retroreflectors and rhomboid prisms. These prisms are offered in custom design with reasonable cost.

CeNing’s prisms are able to achieve very high flatness and angular accuracy: L/20 flatness and 2 arcsec tolerance. The surface quality can be up to 10-5 S/D. We have an advanced test device to guarantee the specifications. Additional coatings like anti-reflective, high reflective coatings can be applied upon request. CeNing offers precision optical prisms from prototype to volume production with short lead times.

In addition to prisms, CeNing manufactures a wide of range of optics including lenses, mirrors, filters, beamsplitters and waveplates.