SINTRONES at APTA EXPO 2026

SINTRONES Technology will showcase in-vehicle edge AI computing platforms designed for public transit buses, rail rolling stock, and connected transportation infrastructure. The presentation focuses on low-latency onboard vision processing, proximity sensing, and real-time situational awareness operating independently of remote cloud connectivity.

Core Technology and System Architecture
The hardware architecture relocates high-throughput neural inference directly to vehicle sensors. The VBOX-3631 in-vehicle computer utilizes Intel Core Series 3 processors with an integrated neural processing unit to execute fleet management logic, telemetry analysis, and passenger information services on a unified system. For high-bandwidth camera vision, the IBOX-602P-IP66 integrates an NVIDIA Jetson Orin NX system-on-module with dual GMSL-2 serialization interfaces. Its fanless aluminum chassis meets IP66 ingress protection and incorporates rugged M12 locking connectors alongside a 9–60V DC wide-input power design to handle vehicle voltage transients.

Deployment or Implementation
Deployment in municipal transit fleets involves direct integration with vehicle electrical networks and physical chassis environments. The platforms withstand standard shock and thermal profiles through adherence to EN 50155 railway standards and E-Mark automotive certification. Communication with existing onboard control subsystems occurs through integrated CAN FD buses and high-bandwidth GMSL-2 video channels, allowing straightforward installation into both retrofit programs and newly manufactured rolling stock without requiring structural redesign.

Industry Applications
Target sectors encompass municipal bus networks, school transit fleets, commuter railways, and commercial logistics fleets. Technical implementations include continuous pedestrian hazard identification, vehicle blind-spot monitoring, dynamic proximity detection, and automated passenger information rendering. In railway applications, onboard edge execution maintains real-time monitoring of trackside conditions and mechanical subsystems without bandwidth degradation caused by cellular signal drops in tunnels or remote routes.

Operational Impact and Performance Metrics
Executing computer vision algorithms directly at the edge provides deterministic processing response times below the latency thresholds introduced by wireless networks. By eliminating continuous raw video streaming to offboard servers, systems reduce cellular bandwidth consumption while preserving deterministic reaction times critical for proximity warnings. The IBOX-604-G2 platform demonstrates this spatial awareness in a package weighing under 339 grams, combining an NVIDIA Jetson Orin module, a 3-axis accelerometer, and dual GMSL-2 inputs to compute real-time spatial depth and distance metrics under a 10–60V DC operating range.

Kevin Hsu, CEO of SINTRONES, outlined the architectural objective: "Public transportation is moving toward real-time, onboard intelligence, where vehicles need to understand and respond to their surroundings as conditions change. Our focus is to bring AI computing closer to cameras, sensors, and vehicles, combining real-time intelligence with the reliability required for transportation environments."

Edited by Evgeny Churilov, Induportals Media - Adapted by AI.

www.sintrones.com