Cost Point Determines AV Acceleration, Says LeddarTech

By Louis Bedigian

---

12th October 2018

LiDAR remains one of the key technologies used in the development of self-driving vehicles. There are dozens of companies that have entered the space, hoping to produce the perfect solution for full autonomy.

This has proven to be more challenging than expected, particularly for the automakers that had planned to deploy driverless cars in the near future. At least one manufacturer changed its strategy after realizing that the costs far exceeded what the consumer passenger market was willing to accept.

“We’re familiar with one example of an OEM, which I can’t name, that was evaluating LiDAR for passenger car applications,” said Michael Poulin, general manager of LeddarTech’s Automotive Solutions business. “The company basically took a different direction and tried to aim for ride-hailing as a first application. This is because the costs were prohibitive for rolling out a passenger car with the high levels of autonomous driving that they were targeting.”

LiDAR was not the sole problem. The automaker evaluated all of the materials and components involved and found them to be too expensive to deploy. Poulin estimates that in order for a carmaker to manufacture AVs for individual owners, the complete LiDAR system would have to sell for “well below” $500 per vehicle. He believes that ride-hailing AVs could support more expensive equipment, possibly in the low thousand-dollar range.

“If you look more toward fully autonomous vehicles, there’s some very expensive LiDAR either in prototype or, in some cases, early production stage,” said Poulin. “They are more mechanical and very expensive. The next breakthrough or step forward that we’ll see is a significant cost reduction, with solid-state LiDAR solutions reaching the required performance level.”

In this respect, automakers are not simply looking for lower-cost solutions – they also want to know that the technology can be relied upon when it hits the road. “They want automotive-grade products with functional safety integrated into the design,” Poulin continued. “That will come as the following step. Really what we see is bringing LiDAR to a mass production state, whereas now the only off-the-shelf available LiDAR are only suitable for low-volume applications because of the cost and reliability.”

Going the distance

Visibility continues to be a chief concern among AV developers, so LiDAR manufacturers are focused on increasing the distance and detection of the most difficult objects. For the sides, corners and rear of a vehicle, LiDAR should be able to identify objects or obstacles that are at least 30 m (98 ft) to 50 m (165 ft) away, Poulin said. The demands for forward-facing LiDAR are much more significant, however.

“Ideally, a vehicle should be well above 150 meters,” said Poulin. “The minimum target today is somewhere between 100 and 150 meters, depending on if it’s slower speed or full-speed application. This is the first step that I’d say developers are ready to work with. Longer-term, they’re going to want this to increase to 200 or 250 meters.”

In low light and bad weather situations, the car could still be unable to see what’s ahead. “Heavy snow, rain or fog can impact different capabilities of sensors to properly understand its environment,” Poulin explained. “One way of managing that is if the vehicle is able to detect those conditions, then it can adapt its behavior. For example, it can reduce its speed, just as a human would. From our view it should be quite acceptable that passengers of these vehicles would tolerate or understand this limitation, where the vehicle would be driving slower in such condition. Personally, I’m not sure I would be comfortable in a vehicle while driving at 100mph  in a snowstorm.”

Smarter cities

LiDAR is often associated with self-driving cars but it could be quite useful in building smarter, more intelligent cities. Poulin said that in a busy area, such as an intersection, LiDAR could provide additional assistance and insights.

“With vehicle to infrastructure communication, there can be some additional measures for vehicles to ensure the safety of other road users and people in the city,” said Poulin. “That’s one possible application. There are also many applications for LiDAR for monitoring traffic, tolling systems, speed enforcement or traffic light enforcement.”

Looking ahead to where LiDAR is going next, Poulin said that the main challenges are range, resolution and optimization of cost and form factor. He said that automakers want the technology to get smaller and cheaper while increasing its performance.

“If you look at things like traffic jam highway autopilot, there are applications for which current LiDAR technology is already suitable,” he added. “The main milestones that will be achieved towards deployment of those LiDAR is, in terms of industry elevation and deployment, getting into the manufacturing phase of the system. And from our point of view it’s really about engineering, optimization and design from manufacturing and bringing it into production.”