In 1994, in an interview to the Rolling Stones magazine, Apple co-founder Steve Jobs said, “It’s not a faith in technology. It’s faith in people.” Of course, he wasn’t alluding to self-driving vehicles; he was referring to the relationship that human beings have with technology. In the conventional sense, the Human in The Loop (HTL) harnesses technology for a variety of purposes. But, when it comes to driver-less cars, the relationship between human beings and technology changes. Humans have to trust technology a lot more since they are not the ones in control in this situation.
The race to develop self-driven vehicles has led to the deployment of three leading technologies in the autonomous mobility sector: optical (camera suites), LiDAR (Light Detection and Ranging), and radar. In most cases, these technologies are deployed in conjunction to complement each other in different driving and light conditions.
While a road occupied by robotic vehicles is not a real scenario yet, modern vehicles do contain multiple semi-autonomous features, such as assisted parking, lane change assist, active brake assist, and pedestrian warning system, to name a few. At present, most countries have framed regulations that allow vehicles with up to Level 2 automation (see chart on Levels of Autonomy) ply on the road, except Japan, which allows up to Level 3.
The larger point, however, is that for building fully autonomous vehicles, integration of multiple technologies will be required. Industry trends indicate that LiDAR has emerged as the key technology for making autonomous and semi-autonomous vehicles a reality.
“Since LiDAR is used in both ADAS (Advanced Driver Assistance Systems) and autonomous vehicles, its sensors support precise, reliable navigation in real-time autonomous operation in urban and highway environments. They can detect and track vehicles, pedestrians, and other obstructions to help vehicles safely navigate at various speeds. This includes traveling night and day in a range of road conditions such as rain, sleet, and snow,” says Sinclair Vass, Chief Commercial Officer, Velodyne.
Though the concept of autonomous vehicles can be traced back to Leonardo Da Vinci’s sketch of a three-wheeled driverless cart in 1478, the quest for attaining the holy grail of self-driving vehicles has more recent origins.
Automotive brands like Tesla, Volvo, BMW, General Motors, and Mercedes-Benz, among others, have been betting big on autonomous mobility. In fact, when Tesla introduced driver assistance systems, which the company called ‘autopilot’, it triggered a mini ‘arms race’ towards the ultimate goal of self-driving cars.
So, what is the difference between semi-autonomous and fully autonomous vehicles? A fully autonomous vehicle can drive itself without human intervention. On the other hand, in semi-autonomous vehicles, the primary responsibility of driving the vehicle remains with the driver, who is helped by an array of ADAS features.
The Society of Automotive Engineers (SAE), a global body that sets automobile standards, has defined six levels of autonomy, from 0 to 5, with the help of a visual chart, J3016, which was released in 2018. This chart has now become the industry standard for autonomous mobility and a reference framework for governments to enact regulations. At the highest level of automation, Level 5, no human involvement will be required.
However, there are multiple challenges when we talk about the levels of autonomous mobility. “Getting from Level 3 to Level 5 is an enormous jump because there is no driver to act as a safety net. The car is literally on its own and must be able to react to all situations that might arise. Seeing, perceiving, and acting — at par with humans or better — are three big obstacles towards the implementation of Level 5 autonomous vehicles. The requirement of a legal framework and regulations is one of the most important requirements for deployment of autonomous vehicles,” says Dr. Michael Richter, Commercial Managing Director, Scantinel.
It was in November 2019 that Waymo One became the first ride-hailing/commercial taxi service to offer autonomous vehicles, which operated without a driver in the car, in the Phoenix metropolitan area in the US. The service was, however, paused in March 2020 due to the COVID-19 pandemic. The company resumed service in June 2020.
The LiDAR technology has been around for almost 50 years. Its development started in the early 1960s, shortly after the invention of the laser. However, it was not until the turn of the millennium that LiDAR technology was used by the automotive sector. Motorsports, like many other automotive technologies, were the testbed for LiDAR application as well in the early days.
In 2005, the Stanford Racing Team of Stanford University collaborated with Volkswagen’s Electronics Research Laboratory (ERL) to develop an autonomous car called Stanley, which was a VW Touareg outfitted with five roof-mounted SICK LiDAR sensors, military-grade GPS, gyroscope, accelerometers, and forward-facing cameras.
The prototype went on to win the 2005 Grand DARPA Challenge race that used to be organized by the Defense Advanced Research Project Agency (DARPA), establishing the credibility of LiDAR application along with a host of other autonomous technologies. In fact, the DARPA Challenge and the Urban Challenge competitions are credited with jump-starting the race for autonomous mobility. Today, automobile manufacturers are leveraging LiDAR as a key component in the race to develop safe, autonomous and semi-autonomous vehicles.
LiDAR systems can ‘see’ things beyond human eyes, with range extending from 250 to 500 meters. An array of LiDAR sensors provides a 360-degree view to the vehicle’s computer, which can then activate the necessary response, depending on the situation.
In the automotive industry, radar has long been utilized to control speed, emergency braking, and warning systems. But now, auto-makers have started integrating LiDAR into ADAS. Almost every company working towards autonomous mobility is using LiDAR technology, except for Tesla.
“Tesla’s strategy is built around its neural network. Unlike many self-driving car companies, Tesla does not use LiDAR, a more expensive kind of sensor that can see the world in 3D. It relies instead on interpreting scenes by using the neural network algorithm to parse input from its cameras and radar. This is more computationally demanding because the algorithm has to reconstruct a map of its surroundings from the camera feeds rather than relying on sensors that can capture that picture directly,” explains Chris Gerdes, Director, Centre for Automotive Research at Stanford University.
The current generation of LiDAR sensors in automobiles generate high-definition 3D spatial maps that are processed by the onboard computers to detect obstacles, lane markings, traffic signals, and other vehicles in the immediate environment. LiDAR systems can ‘see’ things beyond human eyes, with range extending from 250 to 500 meters. An array of LiDAR sensors provides a 360-degree view to the vehicle’s computer, which can then activate the necessary response, depending on the situation.
“LiDAR-based features play a preventive role in mitigating crashes and accidents by providing advance warning or additional assistance in steering/controlling the vehicle. The increasing government regulations for vehicle safety and increase in adoption of ADAS technology by OEMs are driving the implementation of automotive LiDAR systems,” says Richter.
With LiDAR sensors becoming more powerful and smaller, the technology is gaining wider acceptance in the autonomous mobility sector. Recently, Chinese autonomous delivery vehicle manufacturer Juzhen Data Tech signed a strategic partnership with Ouster Inc, a leading manufacturer of high-precision LiDAR sensors, for buying 1,190 sensors.
Each of its delivery vehicles will be equipped with up to three Ouster OS1 sensors for mapping, navigation, and obstacle avoidance to enable autonomous delivery between industrial hubs. “We chose to work with Ouster because of the high-resolution of its digital LiDAR. Ouster provides reliable perception data for our autonomous vehicles, which is critical when driving on public roads. Furthermore, we believe that Ouster’s scalable, digital approach will allow us to further reduce costs while improving performance over time, providing competitive advantage,” says Gu Zulin, CEO, Juzhen Data Tech.
In June, Swedish carmaker, Volvo, signed a deal with Luminar Technologies, a US-based start-up, for equipping their next-generation electric vehicles with high-level of ADAS for the supply of Iris sensors. On the day the deal was announced, Luminar stocks (LAZR) traded at a high of $22 on Nasdaq before settling at around $15. Further, the company has hiked its revenue guidance for 2021 from $27.5 million to $33 million and has conducted a demonstration of its next-generation Hydra sensor, which will be launched in 2022.
One of the major stumbling blocks for adopting LiDAR technology was the high cost of sensors. But that has dramatically changed over the past few years. Luminar is offering sensors that cost between $500 to $1,000. Similarly, the world’s largest LiDAR manufacturer, Velodyne, managed to drastically reduce the cost of its sensors by shifting to “solid-state” technology. This helped it eliminate the moving parts in the optical system, enabling dramatic reduction in size and weight and aiding in scaling up manufacturing manifold, giving the company economies of scale.
“Past 2025, I think you will see the cost of LiDARs come down even more, in the $700 range per vehicle,” said Anand Gopalan, CEO, Velodyne, to Nikki Asia in a recent interview.
Globally, automobile manufacturers are incorporating LiDAR-based applications in several of their existing models to provide ADAS, but in India the uptake is slow. “India is yet to adopt ADAS. Not much has been done in the mobility sector here when we talk about LiDAR,” says Bharat Lohani, Professor at Indian Institute of Technology (IIT), Kanpur.
In India, the perception is that LiDAR technology is too expensive, despite its wide scale adoption in aerial mapping. “LiDAR being expensive is a misnomer if we consider the cost-benefit ratio. There are capex investments in procuring sensors and platforms, but not much is being done in developing the technology in India. On the data processing side, Indian expertise is quite rich as the majority of LiDAR data of the world is processed in India and because of this reason, the sector attracts investments. However, nothing much is happening towards the creation of HD (high-definition) maps, which are important for ADAS and driverless vehicles. There are interested companies, but they are looking for the adoption of ADAS in India. Globally, the creation of HD maps for ADAS is common,” adds Lohani.
Some Indian manufacturers have started adopting the technology in their cars. “Indian roads have challenges, but models like Tata’s Elxsi and Mahindra XUV 700 come with the LiDAR sensors and this is a step ahead, and slowly we too are getting there. But what is important is, be it autonomous or semi-autonomous, vehicles should know about the surface or ground and that is only possible with sensors, cameras, and radars,” he elaborates.
While achieving 100 percent autonomous mobility is a slow process, and to a great extent depends on the speed at which regulations are enacted, semi-autonomy is here to stay. It’s not a question of if, it’s when the entire automobile sector will go autonomous. And will LiDAR lead the way? Only time will tell.
© Geospatial Media and Communications. All Rights Reserved.
