Light detection and ranging (LiDAR) technology has been around for decades, used in the geosciences for surveying and mapping, in police radar to check if you’re speeding, and more recently in smartphones to create better and more accurate augmented reality experiences.
It might also be the tech at the core of future autonomous driving systems.
LiDAR, in essence, is a kind of super-charged sonar. Lasers are fired, in pulses, at the area being measured. The speed at which those pulses are reflected back to the device emitting the LiDAR shows how far away the thing is and, over time, if and how fast the object is moving.
The secret sauce is really in the software that receives that information and interprets it to say that object is a pedestrian, or a bus, or a brick wall; the object is walking perpendicular to the vehicle, running for the traffic light, is predicted to enter the path of this self-driving car in 5.3 seconds.
Federico Collarte is chief executive and founder of Baraja, an Australian-based company that desires nothing more than to bring this traditionally very expensive LiDAR tech into every vehicle on the road.
“Success for us is our technology having the biggest impact in the world possible,” he says. “And to do that, you need to have something that is yes, high resolution and high performance, but also at a price point that it makes mass adoption possible.”
Baraja is a finalist in two categories of the inaugural InnovationAus Awards for Excellence – Advanced Manufacturing, as well as the Wildcard category. You can reserve your seat at the December 1 black-tie gala here.
Baraja’s LiDAR has already been deployed in some mining settings, where the tech can be used as part of a transition to a fully automated mining operation and to, simultaneously and with a high degree of accuracy, dynamically map the ever-changing landscape of a mine’s footprint.
It’s pretty well agreed, scientifically, that LiDAR maps surroundings so well as to be almost faultless. The stumbling block in its broader use up to this point has mainly been price point. It used to be very, very expensive.
Additionally, LiDAR is not a stand-alone solution for any and all self-driving challenges. It has trouble in fog, for example, because fog reflects light. In those cases, one of another core array of special awareness tech would be better to step in, for example traditional radar, which penetrates fog quite happily.
The main point is what it does do, it does very well. Baraja’s Spectrum-Scan product is designed to balance precision with demands on processing power, making it more efficient, much more cost effective and therefore, broadly adoptable.
“The key thing about our LiDAR is, it is high resolution, high performance, but you don’t have to have high resolution everywhere in the field of view all the time,” Mr Collarte explains.
He likes to use the analogy of the human eye, which has a very high resolution focused on a point in the centre of the field of vision, roughly the size of a human face. But we also have a wide field of vision at a lower resolution and a mobile neck and eye muscles that allow a person to change that focal point to gather more detailed data when needed.
“And we have the eyes tightly coupled to a perception system,” Mr Collarte says, “which is the brain. So, we go from low resolution detection: there is something there, but I don’t know what it is. Then I go high resolution classification: it is a pedestrian moving in this direction. And then I can go back to low resolution tracking, because now that I know that it is a pedestrian, I can predict how the pedestrian will behave. And I just need low resolution tracking to do that. Our LIDAR has that same ability.”
And it is this ability to make highly accurate and most importantly early measurements and then make machine learning driven predictions that Collarte says makes Baraja’s type of LiDAR such a vital component of the suite of functions that will potentially drive the cars of 30 years from now, independently and without human input.
There is a common ethical dilemma discussed amongst technologists and philosophers alike when it comes to autonomous vehicles. It’s called the Trolley Problem. Say you’re in a tram full of people trundling down the road in South Melbourne. You approach a split in the tracks.
The tram is currently heading towards the left fork where three men are working on the track. You can manually change course to take the right fork, but unfortunately one man is also working on that track. You cannot stop the tram in time to save both.
What do you do? Do you run over the three men you’re already headed for? Or do you manually make a choice to switch tracks and instead take the life of the lone worker?
Is it always better to preserve quantity of life? If you change tracks to kill the man alone, are you more morally responsible for his death because you made a conscious decision to kill him?
What if, instead of a lone man, that one person was a child who has run into the street? Does that change your decision?
Now what if instead of a tram, that vehicle is your autonomous car?
In the case of a catastrophic collision, for example, should the vehicle seek to protect its passengers above all else? What if it’s about to crash into a whole crowd of people?
Researchers have conducted focus groups to find out what people think about some of these issues.
The studies found that if you asked people to imagine themselves as the driver in this scenario, they felt the car should protect the passenger.
Perhaps unsurprisingly, when asked to imagine themselves as a pedestrian, people felt the car should protect the pedestrians. It’s a major and ongoing challenge for self-driving software designers.
Mr Collarte believes that LiDAR has a role in incrementally improving the safety of all generations of vehicles between now and then and might even be able to eliminate or at least mitigate the need to resolve a lot of these very cumbersome moral dilemmas.
“LiDAR, like the one that we’re building, which is long range, can give you more time to react, because you’re going to detect that problem far sooner,” he says. “So, you can maybe completely avoid a Trolley Dilemma by, for example, coming to a complete stop.”
Mr Collarte says LiDAR will enable what is called “level four autonomy”, where the driver you can go to sleep, or be reading – paying no attention to the driving process at all.
“But it’s going to be a long road with mixed scenarios between now and then,” he says. “LiDAR will also improve outcomes for safety, when the human driver is in control. If there’s a collision, the car starts priming the brakes, or preparing the airbags, etc.”
These are features that some premium vehicles already have, “LiDAR is just going to take that to the very next level” says Mr Collarte, and hopefully make that level of safety available to everybody at every price point.
Baraja is a finalist in the InnovationAus 2021 Awards for Excellence in both the Advanced Manufacturing category and the Wildcard category.
Do you know more? Contact James Riley via Email.