Artificial-Intelligence Applications for Smart Cities and Smart Homes

On Jan 8^th 2018, the Industrial Technology Research Institute (ITRI), known to be Taiwan's largest and one of the world's prominent high-tech applied research institutions, announced that it will launch and demonstrate artificial-intelligence (AI) applications ideal for smart cities and smart homes, at its booth Tech East, LVCC South Hall 2 - 25611 at CES 2018.

Two of its AI technologies include CES 2018 Innovation Awards Honorees: The Intelligent Vision System for Companion Robots plays Scrabble; the Handheld Pesticide Residue Detector capable of determining if the amount of pesticide residues is safe while washing fruits/vegetables and allows the user to know when washing is complete. DeepLook, another AI application, is a large-scale, cloud-based, video analytics technology, used to incorporate video streaming and computing resources; detecting tailgating, video analysis and identification for tracking stolen vehicles and monitoring doubtful activities.

ITRI’s seven AI applications for smart cities and smart homes at CES 2018 also include: Genki Bot (employs AI in order to optimize a user’s sleep environment) and ITRI MusicAI (categorizes musical genres and creates new music). LIM-3D is a multi-layer 3D antenna design and manufacturing technology. Optical Camera Communications (OCC) technology employs LEDs and a smartphone camera to attain visible light communications, launching a new way of distributing and also displaying information. All technologies are capable of demonstrating ITRI’s innovation capability of providing innovative solutions for smart living.

ITRI’s Intelligent Vision System for Companion Robots (a CES 2018 Innovation Awards Honoree) incorporates AI, 3D vision recognition and eye-hand coordination technologies in order to play Scrabble and study from experience. ITRI’s Intelligent Vision System for Companion Robots is capable of delivering the following technology breakthroughs:

• Intelligent Vision Technology and Developmental Learning Position Detection for perception of objects: In the companion-robot demonstration at CES 2018, the robot is capable of distinguishing between different Scrabble cubes, words and their locations. It can also recognize the words spelled by other players and individual letters. The same intelligence applies in different industrial-automation and Industry 4.0 applications; for example, a robot available with ITRI’s Intelligent Vision System could distinguish objects based on their positions, size and surface details even if their dimensional information is similar.
• Smart Grip Technology for interaction with objects centered on perception: In addition to perceiving varied objects based on shape, color, size and location, ITRI’s Intelligent Vision System for Companion Robots allows a robot to grip, move and then interact with the objects, while preventing collisions with various other objects. In the CES 2018 demonstrations, the robot will be able to drive its arms to grip and then move the Scrabble cubes and position them in correct cells without hitting other cubes or dropping them.
• Deep learning: The Intelligent Vision System for Companion Robots allows a robot to adapt to varying conditions and then act accordingly. In the CES 2018 Scrabble demonstration, the robot examines the random moves of its human opponent and shifts its cubes based on its existing abilities. It then verifies spelling via visual input and the dictionary of its game engine, and selects the spelling that scores highest from potential combinations.

ITRI’s Handheld Pesticide Residue Detector, a CES 2018 Innovation Awards Honoree, is considered to be a portable device that has the potential to determine whether pesticide residues are within a safe range when washing vegetables or fruits. It is also considered to be a breakthrough in pesticide detection for consumer use because until now, pesticide detection was available only via commercial laboratories. The Handheld Pesticide Detector is capable of detecting eight washable pesticides from the top 10 pesticide ranking lists in China and the U.S. If employed in tandem with specialized vegetable and fruit cleansers, it will also be able to detect fat-soluble pesticides up to a 0.5 parts-per-million (ppm) detection limit.

The Handheld Pesticide Residue Detector easily operated and read. The user pushes one button in order to activate the detector, and then places the detector in water with the vegetables or fruits. This is followed by the detector displaying lights on the wireless charger specifying results. A red flashing light specifies danger, meaning additional washing is required; a yellow flashing light is displayed while washing is in progress; and a green flashing light highlights that washing is complete.

The Handheld Pesticide Residue Detector comprises of patented technology innovations that make the affordable, portable, real-time pesticide residue detector possible. It uses a “Dynamic Degradative Algorithm” and a “Fabry-Perot Detection Spectrum System.”

• Fabry-Perot Detection Spectrum System
  ITRI has designed a chip based on the Fabry-Perot spectrum filter, covering multispectral absorption frequency bands of frequently used pesticides, in order to calculate vital pesticide concentrations. Conventional optical grating for beam splitting often needs matching linear array light sensors, such as contact image sensors (CISs) or charge-coupled devices (CCDs), accompanied by higher-cost reading circuits and modules. The Fabry-Perot device for the conduction of beam splitting however needs only one light sensor (for example, a photodiode sensor). This majorly reduces system size and also decreases production cost, while improving the signal-to-noise ratio (SNR) of light signals. Furthermore, the Fabry-Perot component is more durable against collision or vibration than standard systems.

• Dynamic Degradative Algorithm
  The content of the pesticide differs based on washing time for different vegetables and fruits. ITRI’s dynamic degradative algorithm was thus developed for real-time monitoring of the pesticide washing process. The system is capable of automatically deciding whether the pesticide concentration level falls within the safety range. This algorithm efficiently detects interfering sources during the washing process, such as water from varied sources, or the existence of a fruit and vegetable cleanser (for example, baking soda); allowing it to be easily introduced into different household scenarios.

ITRI’s DeepLook Helps Fight Crime

ITRI will also be featuring at CES 2018 its game-changing technology, DeepLook, for city security surveillance. DeepLook is known to be a large-scale cloud-based video analytical technology that incorporates video streaming and computing resources. DeepLook is responsible for monitoring traffic activities so that law enforcement and public safety agencies will be able to rapidly respond to reported incidents. It allows video condensation, license plate recognition, and camera network monitoring. It is also capable of detecting tailgating events or tracking stolen vehicles. It leverages prevailing video surveillance infrastructure and integrates cloud computing capabilities in order to improve flexibility and scalability of incident analysis, and mass-scale video analysis, summarization, and condensation. DeepLook can incorporate in seconds any image or video analytical engines (as well as the GPU supported AI/DNN model), and it also automates and then transforms time- and resource-intensive chores into readily available information.