This time next year an expected 10 million foreign visitors and millions of Japanese from Hokkaido to Okinawa will converge on the Tokyo Metropolis for the Games of the XXXII Olympiad. Local authorities are proceeding with multiple AI-based measures designed to deal with the threat of terrorist attacks and other dangers during the Games.
The Japanese government has been cooperating with tech companies on a vigorous AI development policy to maximize security while minimizing inconvenience and people flow problems. These parties are working with Tokyo Metropolitan Police to deploy intelligent identification systems that can detect terrorists and other dangers. The Tokyo 2020 Olympics and 2020 Paralympics will be the first application of AI-based security on such a large scale.
AI deployment will focus on security precautions and monitoring while compensating for the insufficient number of available local security personnel. Security precautions include real-time monitoring and automatic identification systems, which will mainly be used at transit arrival points and as crowds file into event venues; and prediction systems, which will help improve people flow and transit efficiency at and around event venues.
Prediction Machines — Crowd Forecasting System
In order to improve the security efficiency, Tokyo Metropolitan Police are cooperating with Panasonic on a new crowd forecasting system. Cameras installed on police cars will upload human flow data to the cloud for calculation and analysis to predict the dynamics of future crowd movement. The real-time prediction system can also detect suspicious situations such as retrograde motion, suspicious objects placed among crowds, or vehicles traveling in restricted areas. The system will flag unnatural or suspicious activities as possible indications of terrorism and police will be instructed to investigate.
The crowd forecasting system can also be used to provide real-time management of pedestrian and vehicle flow around venues when arge crowds depart together at the end of an Olympic event. The system will leverage pre-collected and real-time data to access the situation around the venue, predict congestion at each point, forecast flow, and use electronic signs and/or smartphone messages to guide spectators to the best exit route.
Nowhere to Hide — Facial Recognition System
Traditionally, admission to Olympic events or restricted areas is via tickets or ID cards which are manually checked. This system however has both inefficiencies and vulnerabilities. A facial recognition system developed by NEC will be used for identification and authentication at Tokyo 2020 entrance points, the first such deployment of the tech.
The facial recognition system uses NEC’s core biological certification technology and NeoFace Watch solution, which ranks among the best in the world in accuracy. The system will identify the 12,000 athletes along with other staff, volunteers, etc for a total of up to 300,000 people directly related to the Games, confirming identity through dual authentication with facial recognition and ID cards. This system is designed to safeguard against terrorists posing as staff by improperly obtaining ID cards. The system is also expected to ease on-location identity inspectors’ workload and reduce congestion caused by inspection inefficiency.
Additionally, a number of NEC-developed biometric technologies including face, iris, fingerprint, palm print, and voice detection will also be available for identity authentication and other scenarios. Police could use biometric technology for example to quickly determine whether a suspect in custody is a known terrorist.
Robots on Patrol — Automated Aggression Detection
Front-line security monitoring at Tokyo 2020 will see robots replacing the police patrols of previous Olympics. Japanese security company ALSOK’s emotional visualization system will monitor crowds via cameras mounted on squads of autonomous roaming robots. The system does a lot more than identify suspicious packages — it also looks for “the jitters.”
Studies show that people involved in criminal or terrorist activities often display physical signs such as twitches, trembling, flushed complexion, behaviour fluctuations, etc. ALSOK’s AI-powered system measures psychological states based on body signals like the jitters, then assigns values accordingly and colour-codes the people in its environment on a display screen — for example tinting in red any individual whose physical appearance suggests a potentially aggressive mental state.
ALSOK tested its emotional visualization technology at a baggage inspection site for the Ise-Shima G7 Summit in May 2016. Although no terrorists were found, the system could identify impatient and/or irritated people in the queue.
Big Data Forecasting — Heatstroke Prediction and Prevention
The Tokyo Metropolitan Government has also made efforts to reduce risks not related to terrorism. A new system developed in collaboration with Yahoo Japan leverages big data and AI technology to predict heatstroke risks at event venues.
The system uses machine learning to analyze local Yahoo data and heat index (WBGT) information provided by the Ministry of the Environment to predict and reduce the risk of heatstroke in specific geo-units of about 125sqm each. The system will be especially useful in congested areas such as event venues and their environs.
Tokyo 2020 Olympics — A Leap Year for AI
The Olympic Games have been a platform for a number of technological innovations over the years — the 1960 Rome Olympics pioneered live broadcasting, the 1964 Tokyo Olympics saw the debut of the Shinkansen high speed rail, and smartphones became a mainstream live viewing platform at the 2012 London Olympics.
Through cooperation between local government and enterprises, the leading innovation at the Tokyo 2020 Olympic Games will be the deployment of advanced AI technologies on an unprecedented scale for improving security efficiency, solving staff shortages, and providing intelligent early warnings.
Author: Yuu Rirou | Editor: Michael Sarazen