The fallacy of 3D perception: unmasking concepts from costs to possibilities

Couldn’t attend Transform 2022? Check out all the top sessions in our on-demand library now! Look here.

3D perception already plays a vital role in shaping how our cities are built, how our cars are made, and how we interact with the physical world. Smart cities use insights from busy intersections to inform urban planning, manufacturers use 3D sensors to automate logistics, and locations track density limits to improve safety.

But for many people and organizations, this technology is largely misunderstood and as a result ignored, limiting the full potential of 3D perception. The misconception that it is too expensive, with limited usability and “good enough” alternatives has led companies to forgo the implementation of 3D technology.

As more and more companies invest in technology to improve efficiency and bottom line, debunking these misconceptions is the key to unlocking the most powerful tool on the market for insights and automation.

Myth: This is automotive engineering

Most people don’t realize how long 3D sensors have been around. The first light-based experiments date back to the 1930s and began in the aerospace industry when companies used 3D sensor technology to study the atmosphere and measure the height of clouds. 3D radar-based sensors were also heavily used during World War II and were a major factor in the United Kingdom’s ability to survive the Battle of Britain in 1940. Decades later, during the Apollo 15 mission in 1971, 3D sensors were used to trace the surface of the moon.

It was not until the 2000s that 3D technology was implemented in the automotive industry. Today, most 3D sensors are designed and developed for the automotive industry due to the involvement of original equipment manufacturers (OEMs) and the huge amounts of investment in the technology by the automotive industry during the ‘Great AV Hype’ of 2016-2020 .

Unfortunately, this meant that advances in 3D sensors became specialized for one market and the devices became less suitable for other applications. As the automated vehicle market cools and the industry realizes that the road to Level 5 autonomy is much longer and more difficult than expected, the 3D sensor industry is looking to other markets such as retail and security to bridge the gap.

Myth: It’s too expensive

A precious reputation precedes 3D perception technology, which is usually the first deterrent. In fact, the technology has become more affordable. Yet most organizations fail to explore the possibilities of 3D perception based on this preconceived idea. 3D sensors have fallen in price over the past decade thanks to rapid adoption across all industries – even smartphones are now equipped with this hardware.

The fabrication of 3D sensors used to be a highly manual process that required each laser and receiver to be placed by hand by an experienced operator. In recent years, however, the production process has been automated, thanks to the involvement of a number of major car companies, including Lincoln Continental, Bosch and Valeo. In addition, the technology has evolved to include mass-manufactured micro-opto-electromechanical systems or hybrid solid-state devices with VCSEL (Victical Cavity Surface-emitting Laser) lasers, which do not require the manual work of previous iterations.

On the perception side, the computational power required increases as sensors can generate more data, but Moore’s law ensures that the CPUs keep pace. Plus, companies like AMD, Intel, and NVIDIA are working hard to produce computers that are smaller, more powerful, and cheaper with each iteration.

Myth: 2D cameras are good enough

Often considered the cheaper alternative, 2D cameras have been a go-to for most organizations for gaining insights for safety and security, customer journey and productivity. While they are great for surveillance and retrospective viewing, they literally lack a whole dimension of insight. 3D sensors add an extra dimension to transcend the viewpoints seen by 2D cameras. An accurate understanding of proximity is valuable for improving safety and efficiency in many areas of business.

On the other hand, traditional 2D sensors, such as cameras, are passive sensors that require little power and have been around for decades, meaning there are many skilled system integrators available. They also contain color information, an element that most modern active 3D sensors do not provide.

3D sensors such as LiDAR and Radar are not intended to replace 2D sensors; they are meant to enlarge them. In many ways, traditional 2D sensors and 3D sensors complement each other perfectly. 3D sensors provide range information, the ability to work in low-light environments and have been extensively developed compared to 2D alternatives to counteract bad weather. Some 3D perception software on the market today uses deep learning and weather filtering AI to deliver accurate insights regardless of conditions such as snow and rain. The power of 2D sensors comes from the above color information and they often have a much higher resolution, with 4K and 8K being available. Moreover, 2D Computer Vision tools are very advanced and popular among students all over the world.

Many 3D sensor manufacturers are now integrating 2D sensors into their units, which provides built-in 2D/3D calibration and in turn allows each data point supplied to contain both 3D location and color information. The best of both worlds.

So what now?

Data is now the most important factor for organizations and governments that want to make major operational changes. From accounting for expenses to optimizing effectiveness, the more comprehensive a dataset is, the better these decisions will be. Undoubtedly, information from 3D sensors will provide the most useful insights to shape the world of tomorrow.

As we strive to build smarter cities, safer infrastructure and more efficient businesses, understanding 3D perception – and the role it plays in public and private environments – will become critical. In addition to exposing misconceptions, it’s important to continually challenge what technology is capable of, and find new ways to improve and implement the right technology to solve long-standing problems.

Jerone Floor is Vice President of Products & Solutions at Seoul Robotics

DataDecision makers

Welcome to the VentureBeat Community!

DataDecisionMakers is where experts, including the technical people who do data work, can share data-related insights and innovation.

If you want to read about the very latest ideas and up-to-date information, best practices and the future of data and data technology, join DataDecisionMakers.

You might even consider contributing an article yourself!

Read more from DataDecisionMakers

Leave a Comment