TOKYO - Engineers in Japan are developing artificial intelligence-based systems to identify trees at risk of falling or shedding branches, as aging roadside and park trees pose a growing public safety hazard.
A growing number of local governments are considering adopting the technology amid an acute shortage of tree doctors and arborists, who inspect trees and determine whether they should be felled.
The systems currently evaluate zelkova and cherry trees, with ginkgo and other species to be added later.
Between April 2021 and November 2024, a total of 1,732 accidents involving trees in parks and along roads across Japan were reported, of which 110 resulted in injuries or death, according to the Ministry of Land, Infrastructure, Transport and Tourism.
In September 2024, for example, a man was killed when a ginkgo branch broke off and fell onto a pathway in a green space in Hino, western Tokyo.
Trees planted in parks and along roadsides are usually managed by local governments, but they are often inadequately maintained because of a shortage of specialists.
Optim Corp., an information technology company based in Tokyo, and Kyuden Droneservice Co., an operator of unmanned aerial vehicles in Fukuoka, have jointly developed an AI-based system that uses drones to detect dead trees.
Another project is Sumitomo Mitsui Construction Co.'s tree risk assessment AI system, which analyzes photographs of dead or decayed parts of a tree taken with a smartphone or tablet.
Tree AI compares the images with tens of thousands of stored photographs of trunks and branches and, incorporating diagnoses by tree doctors, assesses the risk of a tree falling on a four-level scale.
It produces results within minutes and stores the findings, including the tree's location and diagnostic assessment, in a database.
"The risk of a tree falling increases if it has an internal cavity or mushrooms growing on its trunk. But it's difficult for people without expertise to assess it," said Hidemi Kataoka, a tree doctor involved in the project's development.
Local governments including Kizugawa in Kyoto Prefecture, Miyakonojo in Miyazaki Prefecture and the Tokyo metropolitan government have tested the system. Kizugawa, which manages about 20,000 trees, has been unable to conduct regular inspections because of staff shortages.
"With AI, efficient and consistent inspections are possible," a Kizugawa official said.
About 20 local governments are considering introducing the system, according to Sumitomo Mitsui. The company plans to launch the service commercially in fiscal 2027.
"A large number of trees were planted during Japan's high economic growth era for purposes such as absorbing vehicle emissions and are now reaching the end of their lifespan," said Sho Tago, head of the company's green infrastructure business section.
"We hope this service can support the efficient and proper management of trees with limited personnel," he said.
