class="info-alert">『Your web browser does not support JavaScript, but it does not affect browsing through the rest of the web site.』
jump to main content

Industrial Technology Research Institute

:::

ITRI Introduces Innovations in AI, Robotics, and ICT at CES 2022

Date:2022/01/06

ITRI's tech highlights in AI, robotics, and ICT at CES 2022.
ITRI's tech highlights in AI, robotics, and ICT at CES 2022.

ITRI, Taiwan’s largest and one of the world’s leading high-tech applied research institutions, today announced it will introduce AI, robotics, and ICT technologies at its CES 2022 booth 9513, North Hall, LVCC, and its virtual event site (https://event.itri.org/CES2022).

ITRI’s featured technologies include the RGB-D AI Robot, a CES 2022 Innovation Awards honoree, the first collaborative robot that integrates smart 3D vision as a built-in standard; the Autonomous Selfie Drone, which flies itself to take the best picture based on advanced photography techniques; and the Interactive Time Machine, which automatically creates a unique real-time full-body 3D interactive avatar and experience combining a user’s motion, 3D model, surroundings, and display.

The RGB-D AI Robot, a CES 2022 Innovation Awards honoree, is the world's first collaborative robot featuring 3D vision as a built-in standard. The smart 3D vision sensing technology functions as a pair of sharp eyes on a robotic arm. It reduces the size of the sensor by 168 times compared with other robots, while increasing the sensing speed by 38.6 times. The smart 3D vision sensing technology provides automatic three-dimensional scanning of the object area and assists the robotic arm in determining the correct angle of picking up objects based on the shape and placement of the object. Possible applications of the RGB-D AI Robot include sorting and packaging objects for smart automation in the manufacturing and logistics industries, while it also has potential use in retail or service businesses and healthcare to directly serve customers.

The RGB-D AI Robot also includes self-learning technology to enable fast, high-precision 3D object retrieval, eliminating the need for revising control programs to teach robots to identify objects, which often takes one to two weeks. Whenever a production line is changed or new materials are introduced, the RGD-D AI Robot can automatically scan tens of thousands of training materials, quickly learning the best handling strategy in less than 12 hours. As the robotic arm becomes smarter, it can achieve auto-learning and recognize the pick-up or suction point for any new object. Consequently, engineering costs are greatly reduced, and production efficiency is boosted.

The RGB-D AI Robot has obtained 34 patents in 16 areas and has been transferred to the world’s second-largest collaborative robot brand, Techman Robot, for commercial development, and is expected to reach the market by March 2022.

The Autonomous Selfie Drone frees the user from photography courses; does all the hard composition, angle, and lighting preparation; and flies itself to take the best selfie picture using AI algorithms and data acquired from essential photography techniques. Users who know photography also can follow their individual preferences to set up different close-up angles, photography body ratio, and compositions. The drone can analyze posture and head orientation to automatically adjust its own position and the angle of depression to achieve the best composition. Its relative portrait mode uses smart framing technology to analyze the user and background, creating a natural and unobtrusive match between the subject and the surroundings. Compared with other selfie drones, ITRI’s innovation distinguishes itself with fully automated photography, smart aesthetic composition, and the direct use of the phone camera, which grants superior resolution and removes the need to convert image formats.

ITRI’s Interactive Time Machine is the first and only system that creates with AI a 3D interactive avatar from the user’s full body rather than from the face only; automatically generates a younger-looking avatar including the user’s surroundings; and forms a metaverse for avatars to interact with players in real time. The user can move freely in front of the system, and the 3D model shown on the display creates the corresponding movements simultaneously. The user can interact with a virtual self that looks younger or older, which creates a unique experience that intertwines the real world with the virtual one. Use cases include VR/AR/MR gaming, personal entertainment, teleconferences where people meet and interact with their 3D models, and exhibitions for interactive activities to attract visitors.

Additional AI, robotics, and ICT technologies that ITRI is showcasing include:

• The DNN Video Analysis OS is a front-end AI image recognition application for smart streetlights, enabling instant alerts on surrounding events, such as illegal parking and traffic violations.

• Unlike most processor chips designed with the von Neumann architecture, ITRI's Ultra Low Power AI Accelerator includes the computing-in-memory (CIM) architecture which carries the outstanding computing efficiency of AI chips (>1000 TOPs/W @1 bit). It can handle variant data movements with high efficiency and low power consumption, providing unlimited application in AI environments.

Virtualization-based Fault Tolerance for Critical Online Services prevents system service interruptions caused by hardware damage, network disconnection or power interruption. It uses a fault tolerance backup mechanism to ensure high reliability of system services, with low latency and system performance overhead, and requires no application modifications.

• The 5G O-RAN RIC is a 5G network management system consisting of an open radio access network (O-RAN) based 5G RAN intelligent controller (RIC) and an element management system (EMS). The RIC includes both non-real time (non-RT) and near-real time (near-RT) RICs. The non-RT RIC collects the RAN information and predicts network behavior through machine learning. The near-RT RIC automatically adjusts the radio resource management (RRM) to the RAN and prevents quality of experience (QoE) from being reduced by the predicted results of the non-RT RIC. The EMS orchestrates managed elements (near-RT RIC, O-CU, O-DU, and O-RU).

ITRI’s AI, robotics, and ICT technologies are launching in the United States at CES 2022. ITRI is seeking technology transfers and business collaboration opportunities for the above technologies. Interested partners are welcome to book a meeting session at https://event.itri.org/ces2022/register.

ITRI’s CES 2022 online press kit is available here: https://ces.vporoom.com/ITRI.
Images, technology fact sheets, and videos of all the above technologies are available here.

Media Contact
Vivian Chen
Office of Marketing Communications
+886-3-591-9154
vivian.chen@itri.org.tw



Related Links