The Safety Challenge on Double-Decker Buses

Standing passengers on upper decks shift a bus’s center of gravity, increasing instability during turns and the risk of falls. While seated-only policies are near universal, enforcement methods vary widely. One public transport provider partnered with AAEON to deploy in-vehicle cameras paired with edge AI analytics to automatically reinforce compliance.

YOLO inference at the edge with minimal latency for instant passenger detection.

Must withstand high-vibration, vehicle-powered environments without failure.

Seamless deployment across 70+ buses with minimal configuration required.

AAEON's BOXER-8623AI, a compact fanless embedded AI system powered by the NVIDIA Jetson Orin Nano, was an ideal match. With Super Mode support, it delivers up to 67 TOPS of AI performance, enabling the high inference speed and multi-stream processing the application demanded.

Its wide -15°C to 65°C temperature tolerance and 12V–24V power input range made it fully suited for in-vehicle deployment.

High vibration and shock tolerance protects interior components in moving vehicles.

Delivers data and power over a single cable — no separate power wiring needed. Built-in over-current and short-circuit protection eliminates DC adapters.

At just 180×136×75mm, the BOXER-8623AI wall-mounts with four screws, expediting deployment across large fleets.

Our PoE cameras on the upper deck feed video to the BOXER-8623AI, which runs YOLO models on the NVIDIA Jetson Orin Nano GPU to detect standing passengers. When a safety issue is identified, the system triggers an alert transmitted via a Wi-Fi module to a modem in the driver's cab, enabling immediate action.

Custom BSP Delivered

• Preloaded optimized YOLO inference model

• Deployment-ready OS with NVIDIA JetPack

• Preconfigured libraries for multi-camera inference

• Driver compatibility evaluations for all peripherals

AAEON progressed from prototype to pilot run within weeks.

Estimated reduction in incidents with potential to affect passenger safety.

Real-time automated alerts ensure full seated-passenger compliance across the fleet.

Reduction in time drivers spend verifying passengers are seated before each stop.

The BOXER-8623AI's robust, modular design and AAEON's ongoing support future-proof the fleet, extending operational lifespan and providing a streamlined channel for software updates.