Overview
Various approaches to acquire 3-D visual data based on different physics principles such as time-of-flight (TOF), structured light, stereovision, and interference have been analyzed. The impact that 3-D depth and tactile imaging sensors bring about for various applications, ranging from consumer electronics to human-machine interfaces, have been addressed, in addition to an analysis of how the technology is expected to mature in the short, medium, and long terms. Advancements in 3-D depth sensors for applications such as advanced vehicles, intelligent transportation systems, security systems, interactive game environments, and service robots have been profiled.
Key Strengths and Limitations
The key useful features of 3D are texture, shape, contextual relations to other object. The list of challenges and advantages are discussed here.
(i) Challenges
One of the main challenges for 3-D is that processing large data sets is inherently difficult, leading to problems in segmentation of the image. This gets further compounded by the lack of understanding of 3-D solutions among customers and end users. Complexity of solutions may not be suitable for all applications such as faster-to-use touch probes in simple mechanical pieces of specific test points.
3-D vision equipment are getting commoditized, thereby impacting price and profit margin of companies, though the market participants have been improving their market share. Maintaining cost level of hardware applications/integrating solutions and factory implementation challenges are getting stronger. Continued presence of incumbent technologies in certain application sectors, especially in consumer and automotive, is making market penetration difficult. Consolidation of business for the purpose of buying into market share of smaller companies in 3-D business by larger established companies has resulted in their domination.
(ii) Trends strengthening 3-D
Manufacturers are working on complementary methods to arrive at 3-D data which are often a combination of several methods leading to higher accuracy. Advancements in semiconductor technology processing capabilities especially in CMOS image sensors are resulting not only in cost optimization but also in higher adoption of 3-D image sensors. Emergence of independent software vendors’ system integrators with innovative solutions has facilitated 3D interpretation for analytical applications. Research groups, consortiums, and startups in 3-D computer vision are undertaking innovative developments in 3-D sensors to expand its application span. Higher requirements on functionality in specialized applications with several sectors from consumer electronics to industrial automation are leading the way to growth of 3-D sensors in diverse markets. Significant R&D investment by DARPA has resulted in development of 3-D techniques for military and surveillance applications.
Innovations across Key End-user Market Applications
For industrial automation applications overall, it is expected that 3-D vision systems will find newer and wider application as the impact of automation and requirements for higher functionality and quality enhancement. As such, these high-end vision systems are well positioned to meet the manufacturing needs of the future. For example, a camera made by SICK (Minneapolis, MN) operates at up to 30,000 profiles per second at a resolution of 1,536 X 512 pixels and includes onboard algorithms for measurement. This camera and other 3-D vision products are used in quality assurance of manufactured parts, robot guidance in de-palletizing and bin-picking applications, and tire manufacturing.
Canesta’s 3-D technology used in consumer applications does not get impacted by ambient light conditions or any physical objects with similar texture or color in the close vicinity that blends with the background and do not require a laser. These capabilities enable discriminating gestures, such as hand gestures from moving objects. Both the game software and camera are aware of exact location of the hand in space. In the case of immersion applications or games, as the player moves, the image gets lifted out of a room and inserted in to the screen action on a kind of virtual green screen. Hence, fine-grained 3-D discrimination becomes extremely crucial.
In automotive applications lane-departure detection and warning systems, intelligent airbag deployment, panoramic rear view cameras, parking assist systems, and adaptive cruise control have been recognized as key for driver assistance and driver and passenger safety. 3-D vision sensors enable new possibilities in the automotive market and further integrate such sensors in cars. As process level enhancements are implemented and as increasing number of functionality are integrated on-chip, 3D image sensors are expected to be integrated into more automotive applications as we grow closer to autonomous care large scale commercialization.
In biometrics and surveillance applications such as L-1 Identity solution’s systems employ a 3-D surface scanner that uses a near-infrared laser and structured lighting methods. The key advantage of using near-infrared is that the system works effectively in poor lighting. Up to 20 frames a second can be captured, building a map of the face of about 20,000 data points. The company’s products are used by government departments, including the US Department of Homeland Security and Singapore’s Immigration Service.
Aerospace and defense sector include potential applications of 3-D automatic target recognition technology point to areas that would benefit from further research. The most promising application is low-altitude airborne reconnaissance (unmanned aerial vehicles or helicopters). Most research in this area is focused on two broad areas–fast 3-D sensing technology and algorithm extensions.
Takeaway
With marked decrease in cost and complexity of systems, manufacturers and system developers have become smarter and envision higher functionality for 3-D vision systems in several areas of application. Vision systems do not cost as much as they did a decade ago, and offer far greater performance with rich data at blazing speeds.
With the industry trend moving toward free-form 3-D geometry shapes, the traditional solution of touch trigger probes is limited because the geometry definition of complex free-form surface requires thousands of data points. Using lasers and triangulation techniques, 3-D vision systems can determine the position, shape, and orientation of objects. This kind of real-world point cloud data for automated 3-D position detection is useful in assembly, contour, and seam inspection. Although 3-D measurement and structured lighting are already well-established techniques for quality control, improvements in both technology and software are extending their capabilities and opening up new possibilities of application.
