Thinking outside the (single) box: how Modular LiDAR improves vehicle integration by reducing cost and risk

Autonomous vehicles always turn heads - they’re typified by their science-experiment look on the road: strange-looking warts on the vehicle, some of them spinning. They scream “prototype” and even Waymo’s efforts to integrate their sensors into something that more closely resembles a car still has a big rotating donut on the roof. Looking to the future, carmakers are struggling to meet dual aims: achieving autonomy in a vehicle that doesn’t look like a research project. Of all the sensors required by the autonomous vehicle, LiDAR is perhaps the hardest to neatly integrate into the vehicle. This short read dives into the why, provides some examples of current approaches, and details recent breakthroughs that could alter the calculus for automakers.

What you’ll glean from this article

• Why LiDAR should be located behind the windshield
• The challenges we need to overcome
• Getting to the answer - thinking outside the (single) box
• How to avoid engineering detours, U-turns and cul de sacs

Why is LiDAR still a big spinning donut on the roof? It has to do with its function as an optical sensor. By optical sensor I mean it sees – just like a human does – and therefore needs a clear and unobstructed view of the roadway to work best. Humans get this unobstructed view by virtue of being behind the windshield, protected from stones and grit, with a clear view that is de-fogged and wiped clean in case of snow, rain and grime. This works brilliantly for human eyes – and it works brilliantly for other optical sensors too – justifying the mounting of automotive cameras behind the windshield. This elevated viewing location provides a much better view of the road, which benefits all sensors, whereas optical sensors in the grill or bumper are blocked by the car in front. Lastly, there are machine perception advantages that come from co-locating different sensor types: close alignment makes correlation between sensors significantly easier, and by sharing equivalent views, the LiDAR is able to more simply cross-check and verify what is seen by the camera, a key piece of a redundant safety model.

The top middle of the windscreen is hands-down the best location for LiDAR, higher and further forward than the human driving position - anywhere else is a major compromise

If behind the windshield is such a great location, why isn’t LiDAR integrated there?

Audi put it in the bumper, Volvo just announced they’re putting it in the roof liner, and the robotaxi companies are installing them on the exterior of the vehicle. But why? The answer to this question gets into how LiDAR works. You can think of LiDAR as a 3D camera, where every pixel in the image has a precise measurement of how far away it is. While cameras create an image with the available ambient light, LiDAR accomplishes this by using a laser to point a tightly collimated beam of energy at objects on the road, capturing the energy that reflects from the object (be it road surface, another vehicle, a pedestrian). As the system moves from pixel to pixel, the laser reaches out and reports back the exact distance at each pixel. The core of LiDAR is the laser, and the problem with lasers is they need power – quite a lot of power if you want them to see far (200 meters is the bare minimum for safe operation at highway driving speeds - the further a vehicle can see when it’s moving quickly, the more time there is to spot obstacles, make confident decisions and operate safely).

Systems slated for production in the next few years to support long-range highway driving applications vary in their power consumption, but fall somewhere in the range of 20-40 watts. This is an enormous amount of power that needs to be managed, and all this power results in the generation of a lot of heat. Heat is a problem for electronics – especially behind the windshield – because the windshield soaks in all of the sun’s energy and gets capital-H Hot. Hardware installed behind the windshield needs to be qualified for operation in up to 125 degrees Celsius. This is a challenging requirement even for regular electronics – but for the specialist laser technology used in high-performance LiDAR, it is tantamount to impossible. The real engineering problem is this: situating Hot equipment in a vehicle location that is already very Hot makes it extremely difficult to engineer a solution to keep everything working reliably without baking, delaminating, or resulting in other showstopper long-term reliability problems. Let’s not even get started on the EMC/EMI issues...

Turns out this is hard. So hard in fact that most automakers and tier 1’s have not been able (with existing LiDAR designs) to solve these challenges, so - despite it being the de facto location for cameras - they’ve avoided putting LiDAR behind the windshield altogether. Automakers instead embarked on engineering solutions to manage the power draw and heat dissipation in other areas of the vehicle. In automotive parlance “to manage” nearly always means “to add cost” and in this case means custom engineering to keep a high-end LiDAR happy in the chosen location. These other locations will now of course need their own defogging (it isn’t always hot, sometimes it snows) and their own cleaning – these are optical sensors after all. Carmakers are having to pay twice for capabilities they already have in the windshield.

An opportunity: improve reliability and avoid driving up costs by packaging modular LiDAR in the windshield

Thanks to some new thinking in LiDAR system architecture and some clever windshield design, we’re already on the way to these far more elegant outcomes. The solution lies in breaking the problem into smaller pieces: separating the LiDAR into two different modules that connect the laser light using optic fiber (the same stuff the internet is built upon). By splitting the LiDAR into to two distinct boxes: one that houses the electronics and laser (remember: hot, doesn’t play nice behind the windshield) and the other that houses the optics (think: pieces of glass, no complex or hot electronics to contend with), the LiDAR system can be integrated into any vehicle by situating just the Optical Module behind the windshield. This results in dropping windshield heat dissipation by an order of magnitude to around 2-4 watts - neatly sidestepping any thermal reliability, EMC or cleaning issues. The Laser Module can be packaged deeper inside the vehicle, where batteries – or the high-power central computer of the car – have pre-existing cooling systems that can be utilized. And at its core, this is what the modular LiDAR concept is all about: reducing cost by ensuring maximum elegant re-use of existing systems. The glass LiDAR optics can sit behind the windshield, where they take maximum advantage of the clear line of sight, the cleaning and defrosting come “for free” because they’re already there, and reliability problems from hot electronics disappear. Better still, this modular architecture dramatically reduces the size of the unit – very helpful when trying to share the location with the camera and other sensors like radar.

One key enabling technology is a windshield design that lets the laser pass through the glass, as conventional windshield glass is designed to block, this will degrade performance of any LIDAR sensor mounted behind the glass. In response to this need, however, Wideye(TM) , powered by AGC, have developed total system solutions to enable safe and durable LiDAR-Compatible windshields : "Wideye(TM)'s solution and infra-red transparent glass breakthrough made it possible to fit the LIDAR unit behind the windshield. It then benefits from the wiped area, the unobstructed view and all the advantages provided by using glass as a protector, enabling the 'eyes of the autonomous vehicle' "says Quentin Fraselle, Wideye(TM) CEO

By separating the sensor head from the engine Baraja is able to optimize the performance of each component and allow customers to optimize their in-vehicle placement.

If you are an automaker that has given up on the windshield location because, despite the cost, perception & integration advantages, you haven’t been able to find LiDAR that can live up to the packaging challenge, please contact us. Baraja is at the forefront of modular LiDAR development targeting solutions to exactly these kinds of problems. Baraja is able to reduce the cost of integration through maximum re-use, and avoid expensive engineering excursions into custom mounting systems to keep LiDAR happy in-vehicle. 

In subsequent articles we will explore how we do this in more detail, and explain how as the number of vehicle LiDARs goes up, how Modular LiDAR can provide significant cost amortization benefits by sharing the cost of the central Laser Module across multiple Optical Modules. Read more about Baraja in these recent articles in Forbes and Techcrunch.