Connect the Dots: Low-Cost LiDAR Needs To Be Integrated Into Connected Vehicles

Create: 06/12/2017 - 17:20
smart car LiDAR

LiDAR was once viewed as almost a standalone system that could replace cameras and radar for visual input for autonomous vehicles. Those distinctive rotating coffee cans developed by Velodyne and made famous by Google’s self-driving vehicles were seen to be all that was needed for full 360-degree 3-D area scans, but then something familiar happened: innovation

It turns out that the cost and awkwardness of the single 360-degree Velodyne design gave many developers reason to pause and take a second look at other options. This opened the door for solid-state LiDAR sensors. Instead of $75,000 or more for a single Velodyne system, multiple low-cost solid-state LiDAR elements, costing $250 or less, can be placed around a vehicle at four key locations, to provide 360-degree visibility for $1,000.

Besides the dramatically lower cost, the second most interesting innovation comes in thinking of how to use LiDAR. Instead of seeing it as standalone, it is now viewed as one more sensing element to complement others, such as ambient temperature and the automobile’s own diagnostic systems, as well as external viewing cameras and radar, should the need arise (Figure 1).

Figure 1: On the path to fully autonomous vehicles, LiDAR will be but one of many sensors that need to be more fully integrated to optimize for cost, power consumption and performance. (Image source:

The new forms of LiDAR (light detection and ranging) still work by bouncing pulsed laser beams off the surrounding environment and using the reflected light to determine range. Millions of pulses give a map of the area, down to a couple of centimeters. With the range increasing to 300 meters in a clear environment, the automobile should be able to process the information in time to allow a car moving at 75 mph to react in time to an obstacle up ahead.

Companies like Quanergy have led the drive to lower the cost of LiDAR, announcing its S3 at CES earlier this year, but they aren’t alone. Innoviz is prepping its InnovizOne aftermarket LiDAR and the InnovizPro for direct automotive channels.

While most LiDAR implementations to date have tended to focus on optical phased arrays, IoT solutions providers need to look closely at the technology they choose. Companies like MicroVision and Innoluce use micro electromechanical systems (MEMS) mirrors to control the light projection and determine angle information. Phased arrays have the advantage of no moving parts, but that doesn’t mean MEMS should be discounted. They are a well-proven technology and Innoluce’s implementation employs a patented 1-D (single-axis) approach vs. the classic 2-D, dual axis (Figure 2).

Figure 2. The Innoluce MEMS-based LiDAR uses a single axis of movement (1-D) to make it less sensitive to vibration and temperature, two critical parameters for automotive applications. (Image source: Innoluce)

This approach makes the sensor less sensitive to vibration and temperature than 2-D MEMS versions, while also making it easier to control and manufacture.

IoT Solution Providers Need an Integrated Approach

With the innovation accelerating around LiDAR in its various forms, and all of them adding more data to the gigabytes of data an automobile will soon be generating every day, it’s clear that automobile manufacturers need to directly address the problem of integrating sensors and processing their data. If they don’t, they risk having to deal with non-optimized data sources that will require higher processing horsepower than necessary, with the added cost, programming time and power consumption that goes along with that.

With that in mind, a more integrated approach is needed. IoT solution providers considering automotive applications need to be thinking about:

  • Becoming a one-stop, Tier 1 integrator of secure and optimized sensing solutions for an automobile, from engine control units to LiDAR.
  •  Finding someone who is an integrator and getting onto their roadmap.
  • Look for another opportunity elsewhere, with the assurance that there’s no shortage of uses for IoT + LiDAR. Topological mapping is just one that’s taking off, literally, using drones.

About Author

Patrick Mannion
Patrick Mannion is an independent writer and content consultant who has been working in, studying, and writing about engineering and technology for over 25 years.

Latest Videos