How SafeZone Revolutionizes Bus Safety with Automotive Camera and Vision AI Detection

The Global Problem: Passenger Injuries from Bus Doors

Passenger injuries caused by bus doors are a recurring safety issue worldwide, creating hidden risks in everyday commuting. Incidents often involve elderly passengers or people with limited mobility, who move slowly and may not clear the doorway before doors close. Children are especially vulnerable due to their short height, making them harder for drivers to notice. Belongings such as handbags, backpacks, or clothing straps can also become trapped, leading to dragging accidents. In crowded conditions, doors may shut on arms or legs as multiple passengers board or alight at once.

These dangers are not theoretical—real cases have drawn public attention. Accidents continue to occur globally, with reported cases in countries including the USA, Canada, Japan, South Korea, and the UK. For instance, in Canada, Guelph Transit and a bus driver faced a $1.25 million lawsuit after a woman was seriously injured by closing bus doors in 2022. In Taiwan, bus operators reported an average of four accidents per bus per month before adopting enhanced safety measures. Similar incidents of passengers being caught or dragged by doors are reported worldwide, underscoring a universal safety gap. Even the most attentive drivers cannot always prevent such accidents, as their focus must remain on safely operating the bus in traffic. Together, these cases demonstrate that passenger injuries from bus doors are not isolated mishaps but a persistent global challenge demanding stronger bus safety solutions.

The Challenges of Traditional Bus Door Anti-Pinch Sensors and Systems

To ensure passenger safety, buses have traditionally used several types of anti-pinch systems.

◆ Pneumatic pressure sensors:

These systems detect abnormal resistance when doors close and trigger the doors to reopen. However, since they require direct contact and a certain amount of force to activate, they often fail to detect small items such as clothing straps or bag handles. More critically, thinner and weaker parts of the human body, such as wrists or ankles, are especially prone to being trapped before the system responds.

◆ Infrared sensors:

Use light beams across the doorway to detect obstructions, yet they are prone to false triggers or missed detections due to environmental factors such as sunlight glare, rain, or dust.

◆ Ultrasonic / Radar sensors:

Emit high-frequency sound or radio waves to detect objects near the doors. Their effectiveness varies since materials reflect sound waves differently, and environmental noise can cause inconsistent detection.

◆ Time-delay logic:

Some buses also adopt time-delay logic to keep doors open a few extra seconds, though this provides no real awareness of passenger presence.

While these systems offer basic protection, they remain reactive and susceptible to environmental interference, leaving safety gaps that continue to put passengers at risk.

From Limitations to Innovation: How SafeZone Redefines Passenger Protection

Recognizing the limitations of traditional anti-pinch systems, SafeZone is developed as a vision-AI sensing system that proactively detects hazards and eliminates blind spots conventional systems cannot address.

SafeZone is the first AI-powered vision system designed to enhance bus door safety by applying advanced image recognition. Using one automotive camera module, one ECU, and high-precision vision AI algorithms, it accurately detects passengers and belongings to prevent injuries. SafeZone overcomes the blind spots and limitations of pneumatic anti-pinch mechanisms, providing a wide detection range of 30 × 200 cm.

Figure 1: Detection trigger area is 200 cm x 30 cm.

Key Features of the SafeZone Vision-AI Detection System

1. Vision-AI Detection

Leveraging vision AI and deep learning, SafeZone detects passengers and belongings at bus doors in real time. Unlike conventional systems that respond only after collisions, SafeZone provides predictive protection by accurately identifying people in various conditions, whether carrying umbrellas, wearing backpacks, or partially hidden in crowded settings, ensuring potential hazards are detected before accidents occur.

For drivers, SafeZone removes the load of constantly checking doors, allowing them to stay focused on the road and reducing distractions. With real-time alerts, automatic door stops, and intelligent reopening features, the system effectively prevents pinching accidents. Continuously enhanced through deep learning, SafeZone adapts to complex scenarios and achieves over 95% detection accuracy, setting a new standard for safety and reliability in the industry.

Figure 2: Large crowds, passengers holding umbrellas, and passengers with partially visible limbs can all be detected.

2. Modular Hardware Design

SafeZone's hardware system is engineered to meet stringent automotive-grade standards, ensuring structural durability and long-term stability. By minimizing mechanical components and optimizing the camera housing, the system reduces installation space and cost while improving mechanical robustness and protection against impact. This compact, reliable design enables seamless integration across buses and industrial vehicles.

With its modular architecture, comprising one ECU and one camera module, SafeZone supports flexible deployment on different bus types. Configurations can be easily customized by adjusting the number of cameras and ECUs according to vehicle size and application needs. In addition, the system intelligently adapts supplementary lighting to ambient conditions, ensuring high-quality image recognition and accurate decision-making even in low-light environments, thereby safeguarding safety in all operating scenarios.

Figure 3: The LED module is integrated into the camera module.

3. System Integration through CAN Bus

The SafeZone Vision-AI Anti-Pinch System has been engineered for seamless integration with vehicle door control systems. Each camera captures the passenger boarding and alighting area, and the visual data is processed in real time by an Edge AI device. Through the CAN bus interface, the Edge AI device communicates directly with the door controller, ensuring reliable synchronization between detection results and door operations. This closed-loop design guarantees that the sensing module and the vehicle’s door control system work as a unified safety mechanism rather than as independent components.

When passengers or objects are detected within the trigger area, the system immediately instructs the door controller to prevent closure. SafeZone also safeguards vulnerable body parts such as wrists and ankles that are often difficult for conventional systems to detect. By stopping accidents before they occur, SafeZone delivers an advanced level of safety for passengers and sets a new benchmark for intelligent bus door protection.

Figure 4: Seamless integration with door controller through CAN Bus

Figure 5: Even if the driver closes the door due to a blind spot, the system will stop the door-closing action.

Deployment and Future Applications

In Taiwan, SafeZone has already been deployed on hundreds of buses, providing reliable protection against door-related accidents and ensuring passenger safety. Its proven performance in public transportation demonstrates the system’s ability to deliver active prevention in real-world operations.

Beyond buses, SafeZone is designed to extend its protection to cranes, forklifts, garbage trucks, and other machinery where pinch risks are common, safeguarding workers and reducing injuries. With its high level of integration and adaptability, SafeZone is more than an anti-pinch solution; it serves as a core safety module for smart vehicle upgrades across a wide range of industrial applications.

Figure 6: Versatile deployment of SafeZone in transportation and industrial equipment

Learn more about SafeZone: https://www.otobrite.com/product/safezone-regional-safety-ai-sensing-system