02-02-2018 | | By Paul Whytock
A laser system that presently catches speeding motorists looks set to become a critical design element in making autonomous cars safe.
LiDAR (Light detection and Ranging) refers to systems that use laser light emitted in very fast pulses, some at up to 150,000 pulses/sec, and then records the time it takes for the light to bounce back. LiDAR then uses this data to measure distances and create visual maps of roads, traffic signs, obstacles and the surrounding area.Currently such systems have a range of approximately 1200 metres.
The principle behind LiDAR is easy to explain. It’s all about the reflected light beam. For example the beam of light you see when you use a hand torch is actually the reflected beam returning to your eye. As we know, the speed of light is extraordinarily fast, about 186,000 miles/sec and is currently the basis of laser speed guns the police use to nab speeding motorists.
So it's a very fast system. Light moves at a constant and known speed so a LiDAR system can accurately calculate the distance between itself and the target.
The police particularly like laser-based speed guns because they are extremely accurate and unlike radar-based guns they cannot easily be detected by warning devices that motorists install in their cars.
The only flaw with laser speed guns is they would become confused and subsequently inaccurate by alternative laser light beams from an autonomous car zapping the speed gun. This is important bearing in mind that the autonomous cars would be constantly sending out millions of pulsed laser light beams as they make their way along the road.
This however may be a mute point for a couple of reasons. Firstly autonomous vehicles should be programmed to be speed-safe and legal. Not for them the fun of the traffic light grand prix. Secondly, the laser speed guns used by the police will almost certainly become redundant as they may dangerously interfere with the LiDAR system that is being used on autonomous vehicles and cause traffic hazards.
But before all this becomes a reality it’s a fair question to ask if LiDAR systems are now sufficiently advanced to provide the functions needed to guide an autonomous vehicle safely. And what about the cost of such units which up till now has been discouragingly high?
Velodyne, a LiDAR specialist manufacturing firm claims to have constructed a new device that sees further and in more detail than any LiDAR sensor currently on sale. LiDAR sensors until now worked by firing 64 laser beams in circles. Each beam is separated from the next by an angle of 0.4°, with a range of 1200 metres.
But those specifications aren’t enough to help vehicles in real world situations. For example, a car travelling at 70 miles/hour would have four seconds to respond to an obstacle. Basically, You need a LiDAR that can see further than 1200 metres.
According to Velodyne, it’s new VLS-128 128 sensor provides laser beams with a range of 3000 metres and creates four million data points/sec.
But despite the current high price of LiDAR systems (they can cost up to $100K), the race is on amongst manufacturers of both the actual LiDAR systems and their components to get the automotive companies interested in opening their corporate wallets and placing orders.
Not surprisingly, the Consumer Electronics Show in Las Vegas this year saw several contenders showing their wares.
LeddarTech presented its LeddarCore LCA2 which it claims is the industry’s first 3D solid-state LiDAR IC. The company was not at all reticent about trumpeting the benefits of this new circuit saying it is the world’s first solution enabling rapid industrialisation of 3D solid-state flash LiDARs that meet the automotive industry’s requirements in terms of cost, performance and reliability? Time will tell!
What is certainly true is the LeddarCore ICs generate up to 245,000 digitised waveforms/sec from nearly 1.3 billion samples and these are processed using LeddarTech’s proprietary signal processing algorithms, which perform over 25billion operations/sec.
Also at the Las Vegas event was Mobileye with its EyeQ technology which the company says has already been adopted by many automotive companies.
It is currently working on its fifth generation SoC, the EyeQ5, to act as the vision central computer for fully Autonomous Driving (Level 5) vehicles that could be on our roads by the turn of the decade.
To meet power consumption and performance targets, EyeQ SoCs are designed in VLSI process technology nodes down to 7nm FinFET. The company believes it has been able to achieve the power-performance-cost targets by employing proprietary computation cores (known as accelerators), which are optimised for a variety of computer-vision, signal-processing, and machine-learning tasks.These accelerator cores have been designed specifically to address the needs of the ADAS and autonomous-driving markets.
The EyeQ5, SoC will cope with the processing demands of more than sixteen multi-mega-pixel cameras and other sensors. Its computational power targets 15 trillion operations/sec while drawing only 6 Watts of power in typical applications.
Given the potential global market for this sort of autonomous car enabling technology its not surprising that chipmaker Intel purchased Mobileye for $15.2billion. And Intel is not alone in eyeing up the automotive sector. Qualcomm coughed-up a lavish $47billion to acquire NXP. A move that is clearly aimed at capitalising on the component requirements of existing and future automotive markets.
As for our locally revered traffic police and their laser speed guns, I suspect there must be a technology company out there already working on a system that will actually exploit the in-car LiDAR system data to tell the police exactly what speed an autonomous car is going and where it is. A simple computation could then determine if it is breaking the speed limit in any particular speed-controlled area and a speeding ticket automatically issued.
The real question is who should that ticket be emailed to; the car owner or the company that erroneously programmed the system controlling the autonomous car?