Closed Space: What Happened

A passenger sitting in a seat on a highway bus notices the handle of a knife peeking out from the pocket of the man next to him—at that moment, there is no escape. On a train, a person is awkwardly holding a smartphone in the direction of the passenger—there is barely a few seconds to decide whether to speak up or quietly move away. At a railway crossing, the barriers remain down as the train stops, and an announcement plays inside the carriage—the only thing passengers can do is wait.

In late April 2025, a series of such incidents were reported primarily in Hiroshima Prefecture. A man in his sixties was arrested on the spot for possessing a knife on a highway bus, voyeurism and stalking were reported on the JR Sanyo Line, and a light vehicle was stalled at a railway crossing, causing a temporary halt in train operations. All of these incidents occurred in the “closed space” of public transport.

These events may have coincidentally overlapped in the same week. However, when lined up together, one question arises:—how well do the mechanisms supporting safety in public transport function today? And who do these mechanisms ease the burden for, at what moments?

—

Knife Possession on the Highway Bus—The “Few Minutes” Divided by the Driver’s Report

On April 27, a highway bus stopped on a road in Asaminami Ward, Hiroshima City, and a man in his sixties was arrested on the spot on suspicion of violating the Firearm and Sword Control Law. The man was reportedly carrying a knife in his pocket.

What is noteworthy is that the incident ended in “attempted” rather than “completed” crime. The driver sensed something was wrong and reported it to the police. The bus stopped at the nearest safe location, and the driver maintained order in the bus until the police arrived. Although the time from the report to the arrest has not been disclosed, it is undeniable that this “few minutes” of judgment in the closed space of the bus made a significant difference in the outcome.

Highway buses have structural difficulties compared to route buses and trains. There is no luggage inspection when boarding. There are few stops along the route, and the timing for passengers to disembark on their own is limited. According to the Ministry of Land, Infrastructure, Transport and Tourism, while the installation of security cameras on highway buses is recommended, it is not mandatory. The installation rate varies by operator, and for small and medium-sized operators, the installation costs—ranging from several hundred thousand yen to over a million yen per unit—are said to be a barrier.

In other words, at this point in time, the safety of highway buses largely depends on the “judgment of individual drivers.” Even with security cameras, they serve merely as post-event evidence preservation devices; when a person with a weapon is right in front of you, video footage does not protect passengers. What protected them was the driver’s choice to “report” the incident.

—

Voyeurism and Stalking on the Train—Conditions for Cameras to Act as a “Deterrent”

In the same week, multiple cases of voyeurism and stalking were reported on the JR Sanyo Line. On April 17, a stalking incident occurred on a train, and prior to and following that, cases of suspected voyeurism were confirmed.

The installation of security cameras in railway vehicles has rapidly progressed in recent years. JR West has indicated a policy to standardize the installation of onboard security cameras in all new conventional line vehicles by the fiscal year 2023, and retrofitting in existing vehicles is also being gradually implemented. The interpretation standards of the “Ministerial Ordinance on Technical Standards for Railways,” revised by the Ministry of Land, Infrastructure, Transport and Tourism in 2022, effectively made the installation of security cameras in new vehicles a standard requirement.

However, having cameras and having them function as a “deterrent” are not the same thing. Voyeurism and stalking are acts that are completed in a matter of seconds to tens of seconds. If there is no one monitoring the footage in real-time at the moment of the crime, the cameras remain as mere “recording devices.” The victim must raise their voice, those around them must notice, and the information must be conveyed to conductors or station staff, who can then respond at the next stop—only when this chain of events is established can the damage be minimized.

Conversely, for the presence of cameras to serve as a deterrent, the awareness of being “filmed” must exert enough psychological pressure to dissuade the perpetrator. Whether stickers indicating the presence of cameras are posted inside the train, and how many cameras are installed and positioned to reduce blind spots—such “visualization” designs significantly influence the effectiveness of the devices.

Another often overlooked aspect is whether there is a mechanism for victims to “speak up.” JR West has increased the number of vehicles equipped with emergency reporting devices that allow passengers to communicate directly with crew members, but how many passengers know the locations of these reporting devices? The mechanisms exist. However, the “last few meters” to reach these mechanisms can be the most distant distance for victims.

—

Stalled at the Crossing—Between the Mechanism to “Stop” and the Decision to “Move”

In the same week, a light vehicle stalled at a railway crossing between Iwakuni and Tokuyama on the JR Sanyo Line, causing a temporary halt in train operations. It has been reported that no passengers were injured.

Preventing crossing accidents is one of the oldest challenges in railway safety measures. According to statistics from the Ministry of Land, Infrastructure, Transport and Tourism, there were approximately 200 crossing accidents nationwide in the fiscal year 2023. While this is a significant decrease from the peak of over 3,000 incidents annually in the 1960s, it has not reached zero.

Currently, many crossings are equipped with obstacle detection devices (photoelectric sensors or loop coil detectors). If a vehicle or person is left in the crossing, the sensors detect it and alert the approaching train driver with a “special signal light.” Additionally, emergency buttons are installed near the crossing, which can be pressed by drivers of stalled vehicles or passersby to trigger a signal.

This mechanism is extremely effective at “stopping” trains. However, the decision to “move” the stopped train is made by the train driver and the personnel at the control center. Safety checks, recovery forecasts, and arrangements for alternative transport—all of these rely on human judgment and coordination. In this incident, while the exact duration of the halt has not been disclosed, it is certain that the movement of several hundred to several thousand passengers was temporarily halted.

The mechanism to stop trains can be automated. However, the decision to move them cannot be automated. —This asymmetry is the fundamental difficulty of crossing troubles.

—

A Common Structure Reflected in Three Incidents

A knife on a bus, voyeurism on a train, and a stalled vehicle at a crossing. The nature of these incidents is entirely different. However, when viewed together, a common structure emerges.

The mechanisms have reached the point of “preventing the worst-case scenario.” However, there is still room for improvement in the design that shortens the “time of feeling anxious.”

While waiting for the bus driver to report, passengers were in fear. While the victim of voyeurism on the train made their way to the emergency reporting device, the damage was already occurring. While the train was stopped at the crossing, passengers inside the carriage waited without information.

In each case, people are exposed to anxiety in the “gap before the mechanism operates.” Security cameras, emergency reporting devices, and obstacle detection devices each fulfill their roles. However, the current mechanisms do not sufficiently address the question of what passengers feel and what they can do during the time—the few minutes or seconds—before and after the devices are activated.

The Ministry of Land, Infrastructure, Transport and Tourism is promoting a “comprehensive strategy for ensuring safety and security in public transport” starting in the fiscal year 2024, and some operators have begun pilot experiments with real-time video analysis using AI and reporting systems through smartphone applications. The AI abnormality detection technology for onboard security cameras being advanced by JR East—capable of detecting knives and suspicious movements in real-time and notifying crew members immediately—is one attempt to shorten the “gap time.”

However, merely introducing technology is insufficient. It is equally important to inform passengers of the locations of emergency reporting devices, to quickly communicate “what is happening” through onboard announcements, and to create an atmosphere in the carriage that makes it easier for victims to raise their voices—designing these “points of contact between people and mechanisms” is as crucial as the performance of the devices.

—

The Judgment Behind the Mechanisms

In all three incidents, the mechanisms that prevented the worst outcomes were the “systems” in place. The driver’s means of reporting, the recordings from security cameras, and the obstacle detection devices at crossings. However, what activated those mechanisms were the judgments of one driver, the appeals of one victim, and the act of one passerby pressing a button.

Mechanisms are activated by people. People can act because mechanisms exist. —When this relationship breaks down, the “closed space” of public transport becomes a true closed space.

The series of incidents that occurred in Hiroshima also serves as examples of mechanisms functioning. Therefore, what we should be asking now is not “why couldn’t it be prevented?” but rather, “what is needed to deliver this mechanism a few seconds earlier?” Those few seconds can slightly shorten someone’s fear.