Luminar, for the uninitiated, is a San Francisco-based lidar startup that emerged from stealth in 2017 with $36 million in funding. Then in June this year, the company announced a closer tie-up with Volvo to provide lidar for its burgeoning self-driving vehicle efforts, while Volvo took a stake in Luminar via an investment from its Volvo Cars Tech Fund.
The lidar touch
For autonomous cars and trucks to traverse busy thoroughfares at high speeds, they must be able to identify and understand their environment to avoid collisions, a challenge that becomes even more difficult when you factor in conditions such as poor lighting and bad weather. Lidar technology surveys the environment using laser-powered light, and Luminar’s technology promises a significantly longer range and higher resolution imagery than others on the market.
Luminar actually has several OEM (original equipment manufacturer) partnerships in place, including one with Toyota it entered into last year, but it’s the Volvo collaboration that seems to be getting more of the limelight so far. Indeed, at the time of the duo’s partnership announcement in June, Luminar announced a new software suite which it called a “perception development” kit, which automates the process of annotating data from its lidar sensors. So, for example, if a black car ahead has stalled, Luminar’s lidars can not only detect that, but label that detection accordingly.
Volvo was the first OEM to take advantage of this new software.
At the Automobility LA trade show today, Volvo and Luminar announced further developments around its perception technology, which could have significant ramifications for the inherent safety of self-driving cars.
“Pose estimation,” as it’s known, is a facet of computer vision that tries to understand the position of various points in an object, such as the arms and legs of a human being. In the case of Volvo’s R&D team, Luminar’s lidar has enabled them to demonstrate how autonomous cars can understand a pedestrian’s body language. This, in turn, can help machines predict behavior and intention, for example a person walking by the side of a road waiting to cross.
Such technology could be used to identify that a pedestrian is on their phone, for example, and thus determine that their attention may not be as focused on the road as it should be. It’s just one more visual cue an autonomous car can use to make decisions, such as whether to slow down.
“Autonomous technology will take driving safely to a new level, beyond human limitations,” noted Henrik Green, senior vice president for research and development at Volvo Cars. “This promise to improve safety is why Volvo Cars wants to be a leader in autonomous driving. Ultimately, the technology will also create new benefits for our customers and society as a whole. Luminar shares our ambition in making those benefits a reality, and this new perception technology is an important next step in that process.”
In the fast lane
Today’s news comes hot on the heels of a number of Volvo announcements in the autonomous automotive sphere. A couple of months back, the company announced a new concept electric vehicle to illustrate how it envisions the car of the future, which positions the vehicle more as a mini living space where you can work, sleep, and be entertained on the road.
Last week Volvo revealed that its first commercial self-driving trucks will be used in mining to transport limestone along five kilometers of roads and tunnels, representing a milestone moment for autonomous vehicles in industry.
The lidar market is estimated to be a $800 million industry in 2018, and this is expected to rise sharply to $1.8 billion by 2023. And that is why we’ve seen significant investment poured into lidar technology, with the likes of San Francisco-based AEye’s recently raising $40 million for a sensor that meshes camera and lidar data, while others in the space such as Israel’s Innoviz Technologies have also been raising big bucks.
The fledgling autonomous vehicle industry has already experienced its first fatal accident involving a pedestrian, the cause of which was apparently a confused “perception system” that was slow to figure out what the human “object” on the road was, or predict its path. Central to fixing these so-called “edge cases” will be improved technology, including lidar, which will be necessary to take self-driving cars from “mostly” safe to “absolutely” safe.