Open Source Software (OSS) for Sony’s Products
Satoru Ueda has been engaged in a project to widely use Linux for consumer electronics appliances since early 2003. In this project, he has contributed to establishment and enhancement of collaborative relationship between the open source software (OSS) developer communities and Sony. This has promoted active employment of OSS in Sony products, such as TVs, video cameras, and professional visual devices, supporting Sony to manufacture attractive products. At the same time, he has been taking leadership of appropriate use of OSS to prevent legal risks, and continuing support for evolution of Sony-initiated technologies together with OSS communities.
Creation and Productization of New Optical Types That Realize the Reduction in Size and Weight of E-mount Lenses
In order to strengthen the E-mount system and build up Sony's lineup of large-aperture super telephoto lenses and large-aperture wide-angle single-focus lenses for professional camera users in the sports field, Tetsuichiro Okumura has led a project to create small and lightweight optical systems. For the large-aperture super telephoto lens, he exercised unconventional thinking in optical design, and for the wide-angle single-focus lens, he fully took its advantage in new optical system creation. He achieved optical weight reduction by more than 40% while at the same time, improving optical performance in both optical systems. The models equipped with these optical systems have become big hit products, bringing sales far above expectations.
Development of Novel Stacked CMOS Image Sensor with Cu-Cu Direct Bonding
Yoshihisa Kagawa has developed the new chip-stacking technology, called ‘Cu-Cu direct bonding.’ In the stacked CMOS image sensor with Cu-Cu direct bonding, the Cu-connection pads that are fabricated at the surface of each chip electrically and structurally connect the sensor chip and logic chip. This technology has improved both manufacturing productivity and functionality of stacked CMOS image sensor. Such image sensors can be widely used for imaging and sensing purpose. From the R&D phase to the production phase, he has been a leader of process development team and succeeded in the world's first mass production of stacked CMOS image sensors.
Development of Motion Design and Control Framework
Yusuke Kawabe has developed the technology to realize aibo's lively motion and various cute expressions by leading the development of motion design and control framework, and collaboration with other teams. To balance each intended behavior and its natural motion, he designed the motion parts and developed the framework to rearrange the motion. After the launch of aibo, he has further combined new cute and smooth movements to realize the grow of aibo, which makes the users feel reality. This achievement has greatly contributed to the value of Sony’s Robotics business.
Scientific Method for Image Quality Improving
Jun Komachi solved the problems from the dawn of the image sensor technology development, and he made it possible to improve the imaging quality and solve the problem quickly when a manufacturing trouble occurred. He developed the technology combining semiconductor device modeling, measurement and big data analysis jointly with Sony Semiconductor Manufacturing and Sony Computer Science Laboratory. He elucidated the cause of problems that had been difficult to identify, and completed it as a quality improvement method for image sensor products.
Development and Improvement of Pixel-Parallel 14-bit ADC for CMOS Image Sensor
Masaki Sakakibara is the project team leader of the pixel-parallel ADC development aiming to achieve the ultimate structure for image sensors. With his invention, comparator power has been reduced approximately 1/1000 and circuit size 1/50, for an integrated sensor with 3D-stacked pixel-wise Cu-Cu* connection technology. With the effort of his project team, this difficult project succeeded and its result was reported at an international conference. This technology will expand CMOS image sensor business into more new fields.
*A technology to connect electrically by using Cu (copper) pad when stacking backside illuminated CMOS image sensor and logic circuit.
Acoustic Development of the Industry-leading Noise-Cancelling of WH-1000XM3
With the aim to exceed the first-generation model MDR-1000X and deliver the best noise canceling performance for headphones, Shogo Shinkai took the lead in acoustic development, involving developers for creating a new processor. By developing new acoustic structures and signal processing, he was able to control the incoming noise behaviors to the maximum, and achieved the industry-leading noise canceling ratio, which greatly contributed to the competitiveness of the 2018 autumn model, WH-1000XM3. This product has become a hit product, winning various awards in the industry.
3D ToF (Time of Flight) Sensor and System Used in Cars and Mobile Applications
Daniël helped create a new ToF* 3D pixel utilizing CAPD (Current Assisted Photonic Demodulator), and invented circuits to achieve HDR and ambient light resilience in the ToFpixel. Furthermore he developed 3D system algorithms for improving range, accuracy and precision. These achievements enable mobile and automotive 3D sensing solution business for new applications such as gesture control, 3D scanning and navigation.
* ToF(Time-of-Flight) is the method of measuring the distance to the object by detecting the time taken by a light wave from the emitter to travel between the object and the sensor.
Development of the World’s Best Audio Source Separation Method, Its Commercialization, and Boosting of SONY’s Presence
Based on the expertise about the audio signal processing and wide knowledge of the latest machine learning technology, Naoya Takahashi developed an audio source separation method that could almost reach the theoretical limit. Through collaboration with R&D members in Germany and software engineers in India, he achieved the best result in the international audio source separation challenge, Signal Separation Evaluation Campaign (SiSEC), three times in a row. He also published five papers, and presented his work at international conferences as an invited speaker, which has boosted SONY’s presence in academia.
World’s Fastest Distributed Deep Learning Speeds
Large-scale distributed deep learning is an obvious course to effectively reduce DNN training time, but the performance is not stable with a massive GPU cluster. Yoshiki Tanaka and his team addressed 2 novel methods of converging training accuracy in large batch size and low-latency all-reduce for gradient sync among more than 2000 GPUs. They have achieved the fastest deep learning speeds of ImageNet/ResNet-50 training in the world with ABCI (AI Bridging Cloud Infrastructure).
Contribution to Start Licensing of New Porous Carbon Material “Triporous™” Derived from Rice Husks
Seiichiro Tabata developed the new porous carbon material Triporous™ derived from rice husks, in the course of R&D work on battery electrode materials. With its unique microstructure, Triporous has six times stronger deodorant property than the conventional activated carbon, and absorbs eight times more allergens. It also eliminates more than 99% of viruses in water. Recognized as a revolutionary material that gives a new value to rice husks while more than 100 million tons of them are disposed of worldwide, Triporous was awarded the "Encouragement Award for 21st Century Inventions" in the 2014 National Commendation for Invention in Japan, and contributed from the technology aspect to start licensing of this material.
Contribution to RX100M6 by Developing a New Optical System
Koji Toyoda developed a new zoom lens optical system, realizing small and high resolution lens with the aspherical optical design technology. He also developed the new aligning technology and succeeded in mass-production of high-sensitive small lenses. This newly developed zoom lens is used in RX100M6 and greatly contributed to its sales growth by enhancing its differentiation from competitors' products and reinforcing its product power.
Drastic Reduction of Operation Response Time for Professional Camcorders in All Categories
The professional 4K camcorder is a large-scale complex system, composed of many SoCs and FPGAs, to enable the stable and well-coordinated operation of real-time signal processing for 4K acquisition and transmission at 60Hz and the network streaming using open source software. In order to reduce the operation response time of such professional 4K camcorders, Mr. Kenichiro Nagao took the lead on cross-departmental improvement activities aiming to meet the requirement of professional camera users. He performed not only individual component optimization but also thorough integrated improvement, such as delegating process to another chip or reducing wait between chips, and succeeded in drastic reduction of time from power-on until acquisition start to achieve six times faster. He has fully met the expectations of professional camera users, and largely contributed to the promotion and business development in the professional 4K camcorder domain.
Development of Fundamental Technologies for toio™ and Realization of Its Commercialization
Aiming to deliver products that can inspire creativity of children through playing and learning experiences, Tetsunori Nakayama led the toio™ development project from product planning to mass production. Developing Sony's proprietary technologies, such as the absolute position sensor and robot control algorithm, this project has realized self-position estimation and a user participation platform. He has opened up a new robotic toy platform and new possibilities of Sony's recurring business, contributing to the improvement of Sony's brand.
R&D Behind the World’s First Visible-Light Vertical-Cavity Surface-Emitting Laser Using a Curved Mirror
Tatsushi Hamaguchi has realized the first blue and green surface-emitting laser that is continuously operable at room temperature, using a proprietary design with nitride-based curved mirrors. This design represents a breakthrough in lateral confinement, resulting in the best characteristics reported to date. This outcome, including his papers introduced as articles in science magazines, is highly acclaimed nationally and globally in the academia. This element holds the promise of numerous future applications, such as laser printers, projectors, solid state lighting, optical communications, and biosensors, drawing much attention from various industries as the next-generation light source.
Real-time and AI-based Subject Tracking for Sony Camera Autofocus
Houman contributed to the development of the Real-time Tracking technology used in the recent Alpha and RX series cameras. Real-time Tracking is one of the highly acclaimed technologies in Sony camera technology. His aim was to create a robust tracking algorithm so the users can easily take sharp and in-focus images regardless of the subject’s movements and occlusion. He developed an accurate algorithm that significantly improves users' camera real-time autofocus experience using state of the art AI technology.
Contribution to the Commercialization of BRAVIA by Developing High Image Quality Technology “Object-based Super Resolution”
Kenichiro Hosokawa has developed the Object-based Super Resolution technology through the collaborated efforts with Sony Home Entertainment & Sound Products Inc. and Sony Semiconductor Solutions Corporation. This technology is designed to extract each object's features from video images and express the optimal image quality for each object. The algorithm that makes this possible is developed to be lightweight so that it can be implemented in hardware. Being incorporated into the high-quality image processor X1 Ultimate in Sony's premium BRAVIA MASTER series, this technology has contributed to enhance Sony’s brand value.