Innovation in 3D Imaging System – A key enabler for autonomous vehicle

There are many different types of 3D imaging sensor available in the market which operates with different technology principle and having one thing in common i.e. very few light pulses are emitted before the previously emitted pulse is measured and received.

Machine vision in automotive sector in both developed and developing countries, gaining images with high precision and speed is gaining traction. The range system in automotive is associated with the emergency braking system. In addition, emergency braking system to perform fast and reliable operation is very important. According to the European Commission statistics, approximately 25,000 people met with the road fatalities in 2014. In addition according to the International Energy Agency report, the total number of vehicles in the world is projected to double by 2040. This will increase the risk for more road fatalities in coming years. Thus there is a need of a machine vision system which can precisely identify the target at long distance and high speed.

3D vision system light field and illumination system operating with time-of-flight (ToF) principle frequently uses many thousands of pulses to image a single 3D scene. This process further limits the achievable frame rate and imaging of the moving targets. Existing imaging system use fixed direction light field optics or motorized moving optical systems illumination to acquire 2D/3D data of a scene. Sensors with 120 megapixels could not actually provide a high resolution for 2D/3D scenes in a wide range of viewing directions and distances. Whereas indirect time of flight is implemented with single-photon synchronous detection (SPSD) relying on a modulated sinusoidal optical signal and the measurement of phase difference between outgoing and incoming light and takes into account multiple pixel detectors to calculate the phase difference which further leads to slower analog readout and digital conversion. Also, travel time through the indicators is with the end goal that modulations higher than 50MHz are impractical.

Some devices using light at 1.5 micrometer near- infrared (NIR) wavelength are mainly focused for the long distance identification such as LIDAR or range finder. To acquire a complete frame, the current vision system takes lot of time, especially when the scanned scene is a complete landscape. Thus there is a need for a vision system which can identify long distance object in 3D space with very high speed.

To address the above challenges, Fastree3D a Switzerland based company has developed a next generation 3D vision system to resolve the above mentioned challenges such as cost, speed, distance, ambiguity, and adverse illumination sensitivity and in addition increase price/performance by a magnitude of five to ten times in the field of machine vision, automotive and other advanced vision applications. The 3D vision system developed by Fastree3D will enable to locate objects at long distance in three dimensional spaces with very high speed. This will further enable safe and intelligent systems to operate with high precision and reliability.

There are many different types of 3D imaging sensor available in the market which operates with different technology principle and having one thing in common i.e. very few light pulses are emitted before the previously emitted pulse is measured and received. These sensors are constrained regarding estimation separation and procurement speed. In particular for objects at distance a very limited acquisition speed is implied. In addition, when multiple low power pulses are used over the large time interval, the data throughput is not substantially enhanced because of the lack of on chip capabilities. In addition reception and emission of pulses with the help of common optics causes partial reflections of pulses. To address this issue, multiple low-power pulses is employed which is further detached by large time intervals and to measure a distance for a 3D pixel the total result is averaged out. For obtaining image of fast landscape, this technique is not suitable.

Fastree3D SA is a Switzerland based company located at EPFL campus of Laussanne has improved the current 3D vision system and resolved the challenges related to detection, evaluation and overall cost of the device. Fastree3D has developed a flexible 3D vision system which can be easily integrated with any device using direct time of flight technology and single photon detector with on pixel time digital counter with the help of standard CMOS technology. Fastree3D vision system can be employed in the complex imaging function such as range finding, 3D scanning, position detection, movement detection and linear proximity detector, linear light meter, precision light/curtain barrier and many more. The fundamental highlights of Fastree3D vision framework are high edge rate, low light imaging, programmable locale of intrigue, adaptable pixel binning with individual pixel readout, and programmable control of increase, outline rate and presentation, and specific prohibition of hot pixels.

For different industry verticals such as industrial machine vision, automotive, consumer electronics, robotics and many more 3D vision system is gaining traction for in-depth and accurate information about images and the environment. Availability of in-depth information with high speed is increasing the opportunities to enable better identification and handling of materials in an industrial or warehouse environment. Such in-depth measurement is so accurate that it can separate foreground from the background. Such accurate information about the images or the environment will enable safe and fast operation.

In terms of competition in the field of 3D imaging, Fastree3D novel technology is competing with quite few companies that have already ventured in to this and also trying to take a share in this particular market. So to garner the confidence of the partners and investors is a primary challenge. Fastree3D has overcome the above mentioned challenge by working with the right partners in industrial vision domain that share the Fastree3D vision. The company’s selection of the target market and partnerships has allowed them to build the working samples using their novel technology at very affordable cost levels. In addition selection of target market and its cost effective technology has helped Fastree3D to identify a way to hit the jackpot for future developments. In addition to the above mentioned practices adopted by the company, a team of experts from various fields to collaborate on many different features of the core technology development and a constructive feedback loop are some of the best practices adopted by Fastree3D in developing this novel technology.

Fastree3D will penetrate into ADAS and autonomous driving market to provide next generation of in-depth and accurate imaging. The need to have high quality imaging anywhere such as low and high light intense areas has been met by Fastree3D technology, which enables a wide and diverse range of applications.

Fastree3D intelligent 3D vision system can be easily be deployed in different applications such as machine vision, advanced driver assistance system, autonomous vehicles, personal service robots, gesture recognition and many more. Fastree3D main objective is to open and enable new advanced application in the field of automotive and machine vision. The 3D vision system of Fastree3D will provide unambiguous and precise identification of the moving objects in 3D space at long distance and very high speed. With the help of Fastree3D solution integrated into the device will enable to implement safety features such as automatic emergency braking with better effectiveness. In addition the 3D vision system in long run will be one of the enabling technologies for the development of fully autonomous cars for the mass market. The 3D vision system developed by Fastree3D will drive the overall performance by providing high-quality experience.

Previous PostAutonomous Vehicle
Next PostRechargeable batteries technology for Electric Vehicles