Sydney, Australia-based Baraja will engage in a major hiring spree this year after it closed a US$32 million (just under A$45 million) round of venture capital financing in mid-December.
The start-up, which is developing machine perception technology to help enable the development of self-driving cars, will be adding about 50 staff to its current complement of 95.
“We’re hiring all kinds of weird engineers,” Federico Collarte, the CEO of the company, tells FreightWaves over coffee.
“Electrical engineers, mechanical engineers, mechatronic [robotics] engineers, optical engineers, manufacturing engineers, process engineers… all kinds,” he says.
Collarte is one of two telecommunications equipment engineers who founded Baraja. The other is Cibby Pulikkaseril. Peru-native Collarte explains that the name of the company “Baraja” means “deck of cards” in Spanish.
It also means “to shuffle a deck of cards.” And that’s a clue to the nature of Baraja’s technology.
The company “shuffles” light to help vehicles of all descriptions to perceive the world.
A lidar-enabled people-carrier
Collarte and one of his engineers, Jackson Gritching, give a lift to your FreightWaves correspondent.
Stepping into the back of the air-conditioned people carrier was a welcome relief from the ambient 95 degree Fahrenheit (35 degrees Celsius) heat. It’s hot in Australia right now.
The people-carrier is equipped with a fully functioning light detection and ranging (LIDAR, or lidar) system.
On the top of the people-carrier is a roof rack and, attached to that, are four angled boxes each with one side covered in glass.
They’re laser emitters/sensors for the lidar system. And here’s what’s unusual about them – they don’t move.
Conventional lidar systems have a problem. Lidar uses lasers to scan. As readers may recall from their school days, lasers are very narrow beams of light that travel in one direction. It’s one of the reasons that they are so good for scanning barcodes on groceries.
But for scanning the real-time environment in all directions and all angles? Then that uni-directional nature becomes a bit of a drawback.
Conventional lidar systems work around this problem by introducing an element of mechanical rotation into their systems. Readers may well have seen various cars with funky rotating equipment on the roof on roads near their homes and workplaces. Those vehicles could well have been testing lidar, or a variety of other sensory systems.
Either the whole system physically rotates or the laser is directed onto an angled rotating mirror, which sends the beam out in all directions, to enable the scanning of a three dimensional environment with a uni-directional laser.
A different technology
Baraja’s technology works differently. Baraja installs a prism inside a sensor on top of a vehicle, and then shines a beam of laser light through the prism.
A rainbow of colors then exits from the other side, just like in the picture. What human beings perceive as “light” is, in reality, just a variety of frequencies on the electromagnetic spectrum that the human eye happens to be able to perceive. Any given ‘color’ is just a particular wavelength of light to which human brains assign the concept of a particular ‘color.’
The key thing, from Baraja’s viewpoint, is that light moves forward in a straight line – until it enters a prism. At that point, the prism splits the light into its component wavelengths (or, if you prefer, into its different colors). It also causes each separate wavelength/color of light to change direction compared to the original beam of light.
Different wavelengths/colors of light exit at different angles to each other. The angle of exit is dependent on the wavelength/color of light. Which is how prisms create rainbows from beams of white light.
If an engineer can alter, at super-fast speeds, both the angle and the wavelength of the laser light that enters the prism, then the different wavelengths / colors of light that come out of the prism will be at different angles to each other at that point in time and also to the light that exited the prism a micro-second ago. By changing the angle of the entry beam, a spectrum of different wavelengths / colors of light can continuously sweep from side to side (or, indeed, up and down).
Having changed direction in the prism, the individual wavelengths of light then exit the prism and travel in a straight line – until they hit an object. The light then bounces off the object and returns, also in a straight line, to the prism, where it is detected and transmitted by fiber-optic cable to the vehicle’s computer.
Because it’s light, it all happens at a constant speed, which is, of course, very fast. And because light travels at a constant speed, the computer in the system can count how long the pulses of light take to leave the prism and come back. That way, the computer can determine the distance between the sensor and whatever object the beam of light has hit. Baraja also shifts the frequency of its lidar light a few notches so that it cannot be perceived by human eyes.
And that’s how the Baraja lidar spectrum-scanner works.
Old science… new advantages
There are a few big advantages to Baraja’s scanner compared to other lidar technologies. The main body of other lidar systems are mounted externally, so they are exposed to the weather and impact shock. As they have many moving parts, they are vulnerable to vibration-induced mechanical failure. And, if they get damaged, they’re expensive to replace too.
In contrast, Baraja’s system is solid-state – it has no moving parts. It’s therefore far less vulnerable to all the shocks that vibration, motoring and nature can throw at it. So it should be far more reliable.
Baraja’s lidar scanner is also complementary to other ‘perceiving’ technologies – it can be used alongside cameras, radar and ultrasonic equipment. Radar, for instance, will excel in extremely bad weather situations such as heavy snow and ice that might pose a challenge for lidar. Whereas lidar, unlike radar, will produce an image with a comparatively detailed resolution.
Mass produced components
Collarte points out that all of Baraja’s components – computers, lasers, fiber optics, optical glass and so on, are not only commercially available, they are produced en-mass by reputable manufacturers in the telecommunications industry. So the components of the system are cheap and reliable, especially the roof-mounted sensors which are just “cheap glass.”
The really critical part of the system is the ‘Baraja engine’ which is the tunable laser and the computer housed in a box. It is deep within the vehicle and is “well-protected.” And, Collarte adds, the system is wholly modular – everything other than the Baraja engine itself is off-the-shelf. Everything plugs together. If any part of it gets damaged then it can easily be replaced.
As Collarte explains the system to FreightWaves, it seems like a rather obvious idea. Collarte thinks so too.
“I thought, ‘this is so simple! Why is no-one else doing it? Why?’ It’s a really simple idea,” Collarte tells FreightWaves.
He’s right. Conceptually, it is pretty simple. The basic science of manipulating light with glass is pretty old. The 13th century English monk and scholar, Roger Bacon, theorized that rainbows were created by a natural process similar to passing light through a glass bead. Jump forward a few centuries and Isaac Newton was the first scientist to write extensively in English about light, prisms and refraction.
“The difficulty,” Collarte tells FreightWaves, “is that it is simple to think of, but very difficult to do. You need absolute mastery and control of the laser with millions of changes of points and angles a second.”
Seeing the images in motion
So what does it all look like in motion? Collarte and Gritching have the sensors on the roof of the people-carrier plugged into the Baraja engine. Data from the Baraja engine feeds into a large visual display unit in the back of their people-carrier.
To begin with, the display is a mass of confusing purple and yellow dots. Confusing, that is, until the driver gets the people-carrier moving. The effect is a bit like looking at an optical illusion – all of sudden the image ‘flips’ and the brain correctly perceives what its seeing. Moving dot-point images of cars, trucks, shops and even trees – right down to the bigger branches – suddenly appear. Cars drive by and we can see the shape of the wheels and even the void spaces where the windows are.
“Commodore, Ford, Volkswagen,” Gritching intones the names of each brand of car as it passes by on the screen. He’s not wrong. And he’s not looking out the window either. An unusually shaped truck with a barrel-body drives virtually across the screen and physically in front of our vehicle as we pause before exiting from a minor road. The truck is a cement carrier. And it is as utterly recognizable on-screen just as it is in real-life.
Of course, this is just a mere, puny, human’s perception of what the computer is ‘seeing.’ But, in reality, the computer is perceiving much more and much faster. It sees in 360 degrees. It sees a long way down the road. It sees at multiple vertical and horizontal angles. It knows exactly how far away the lamp-post is from the curb, how far the curb is from that truck that is beginning to overtake us and which your correspondent has only just noticed. And it knows it instantly. And continuously.
We pull up at the CSIRO’s institute where Collarte explains some of the commercial aspects of Baraja.
Sequoia Capital China and Blackbird Ventures (an Australian fund) invested in the recent venture capital funding round, alongside Main Sequence Ventures. Main Sequence is the investment arm of Australia’s national science agency, CSIRO. Sequoia and Blackbird both took seats on the board although Main Sequence did not. FreightWaves sought further information on the structure of the deal, but Collarte declined to disclose details. However, he did comment that “venture capital in Australia is more competitive now,” in respect of its terms compared to previous years. He also disclosed that there is equity for the employees and that the lawyers used by Baraja were the Australian law firm LegalVision. He doesn’t foresee a need for a further fundraising as the company is “comfortable” with the amount recently raised.
The company also previously received, in November 2016, a A$1m grant from the Australian Federal Government’s Accelerating Commercialisation Fund. And, according to the Crunchbase database, Baraja also received funding in a previous 2016 financing round from Blackbird Ventures.
Looking to the near future, Collarte forecasts that self-driving technologies will reach an inflection point in September or October of this year.
When pushed by FreightWaves as to why he foresees that to happen later this year, he commented that the industry had been “very bullish” in the recent past but that self-driving technologies were not ready. It’s his feeling about the market, he says, based on “intimately” understanding the technology, discussing with customers and reading articles, that the inflection point will come this year. Even though, he adds, self-driving cars do not yet exist. It’s Collarte’s view that self-driving cars will be created but it will take along time before it happens.
“Maybe 20 years,” he says.
Baraja is already selling some lidar kits around the world. Collarte declines to disclose details, citing commercial confidentiality. However, he says that kits are sold to “all kinds of land vehicles.”
“On trucks you would have the [Baraja] engine installed on the cab, with fiber optic cables to sensors on the trailer,” he explains, adding that the cables can “easily” be 100 meters long and so could be deployed on road trains and mining equipment.
The ultimate aim of the company is to have its technology embedded in consumer cars. Collarte reports that the company is targeting the mass market because that will enable the company to generate a sufficiently low price point that it could target other markets.
The company is in discussions now with manufacturers, although he points out that car design cycles can be up to seven years in duration for the major manufacturers. Collarte declines to name names on the grounds of commercial confidentiality.
“It has to be really cheap to eventually put it in a car. That’s our advantage. We have mass-produced components.
“We can really win here,” Collarte says.