- One onboard unit takes over ADAS, DMS, BSD, AVM and third-party applications — fewer boxes, fewer cables, one supplier to answer for the whole installation.
- Safety warnings reach the driver first; only events that cross the line reach the depot, with video attached.
- Passenger counts and origin–destination flows arrive from every route, so the network is planned from measured demand instead of assumptions.
- Data leaves the bus over open, documented interfaces, so agency systems and approved third-party apps plug in without custom middleware.
PUBLIC TRANSIT
Smart Public Transit Safety and Operations
Factory-built onboard AI and open data that make buses safer and networks cheaper to run.
What changes when you deploy
Scenarios on this page
Driving Safety on Urban Routes: city-tuned ADAS and driver monitoring
Collision Mitigation: warn first, assist only when the driver cannot react
Passenger Flow and OD Analytics: counts, load maps and origin–destination insight
Bus-Lane Enforcement with Onboard ANPR: the fleet polices the lane it runs in
Integrated Onboard Control Unit: one box for perception, applications and data out
Operations and Dispatch: live control, duty records and street data back to planning
Transit runs on precision and public trust, and both are built on what the bus itself knows. SOHANIoT manufactures that layer — AI mobile DVRs, ADAS-DMS cameras and onboard control units — in our own facility, so agencies buy hardware that is certified, engineered for the vibration and electrical conditions of daily bus service, and backed by the factory that built it.
Every module on this page is written around the same architectural bet: reduce fragmentation on the bus and in the back office. When the perception systems, the applications and the data interfaces live in one open stack, integration cost falls, and the ridership and operations data they produce starts to drive the network plan itself.
That second part matters as much as the first. A transit network is a system with thousands of moving inputs — traffic, weather, events, passenger demand — and the agency that sees them in one data feed is the agency that can respond before the delay compounds. This stack is designed to be that feed: open, documented and continuous.
Hardware Behind These Modules
Every capability on this page starts with hardware SOHANIoT designs and builds itself. That is why the stack behaves like one product instead of an assembly of parts from five vendors.
- SOHANIoT AI Mobile DVR — the 4G onboard recorder and edge-AI host: multi-camera input, GNSS, impact sensing and on-device ADAS/DMS inference in one certified unit. It records and analyzes in the same box, so there is no separate "camera system" and "safety computer" to reconcile.
- SOHANIoT ADAS-DMS Camera — the driver-facing and road-facing perception pair that feeds the safety features on this page. With the right forward lens, the same camera set also serves ANPR capture.
- Onboard control unit — consolidates the perception feeds and hosts the open application layer described in the scenarios below. The name is a working placeholder until the official product line name is confirmed.
- Vehicle sensors and peripherals — door contacts, passenger counters and optional brake-interface modules connect to the same unit, so one installation covers safety, ridership and enforcement.
Who This Is For
The open-stack design is not an engineering preference; it serves three distinct buyers, each with a different reason to want fewer locked boxes on the bus.
- Transit agencies get the network-planning asset: ridership and operations data they own, in formats they can route into their own planning tools. The loop between measured demand and the published timetable stays inside the agency.
- Bus operators get safer depots and a defensible operations record. Safety events, duty records and service evidence live in one platform, so reporting to the authority does not require assembling files from several vendors.
- Integrators and transit-tech vendors get an onboard platform to build on. Their applications run on the unit or read its data through documented interfaces — no hardware lock-in, no per-fleet bespoke integration.
01
Driving Safety on Urban Routes
A city driver reads a hundred hazards an hour; the system's job is to surface the one that matters.
Calibrated for bus work, not test tracks
Detection profiles are tuned for urban operation: stop-and-go traffic, boarding crowds, turns across cycle lanes and tight terminal loops. Nuisance alerts are filtered, so the warnings that reach drivers stay credible.
Two eyes, one unit
The driver-facing camera tracks fatigue and distraction while the road-facing camera watches for conflicts with pedestrians and two-wheelers ahead. Both run on the same SOHANIoT device, with no extra boxes in the cab.
Around the bus in a single picture
The exterior cameras merge into a top-down view for low-speed maneuvering, terminal turns and docking. Pulling away from a crowded stop, the driver sees the full perimeter instead of assembling it from three mirrors.
Images that hold in tunnels and rain
Wide-dynamic-range cameras keep the picture together in backlit tunnels, night and heavy weather. That is when the warnings matter most — and when most systems fail the driver.
02
Collision Mitigation
Warn first. Assist second — and only when the driver has not acted.
A staged response, not a surprise brake
The system warns of a closing hazard, then monitors the driver's reaction. If the risk persists at the speeds where bus collisions happen, the optional control module can begin a controlled deceleration.
Deceleration written for standing passengers
Intervention profiles are shaped for standees. The bus slows firmly enough to prevent harm, and gently enough not to turn braking into a second incident.
Every intervention explains itself
Each event records the warning timeline, the sensor picture and the video. Agencies review why the system acted, and tune thresholds from evidence rather than opinion.
A layer you switch on when ready
Fleets can run warning-only first and enable controlled braking later. The hardware stays the same; the agency chooses the pace of deployment.
03
Passenger Flow and OD Analytics
A timetable is a guess until you know where passengers actually board and alight.
Counts at every door, every stop
Door-area sensors and AI video count boarding and alighting passengers continuously through the day. Load-per-stop records land in the SOHANIoT Cloud Platform, ready for planning and reporting.
OD patterns without surveys
Matching boarding and alighting events reconstructs origin–destination flows by trip and time of day. Planners see which station pairs actually move people, instead of commissioning expensive manual surveys.
The empty seats become visible
Load maps show where buses run full and where they run empty — by route, direction and hour. Vehicles move from under-used trips to the corridors that need them.
Privacy by design
Passenger data is processed as anonymous counts and aggregates; raw video is never part of the ridership feed.
04
Bus-Lane Enforcement with Onboard ANPR
A blocked bus lane is a delayed timetable — enforce it from the buses already in it.
The fleet is the enforcement network
A forward camera on regular service captures vehicles that misuse the dedicated lane. No roadside gantries to build, power or maintain.
Plates read where they are issued
ANPR is configured for the region's plate formats and encoding, and refreshes over the air as formats evolve.
A review queue, not an auto-fine
Captured events — plate image, time, location and video — are organized in the cloud for the enforcement office. Staff decide; the system prepares the file.
One camera, two jobs
The same forward unit feeds both ADAS functions and lane-enforcement capture. Agencies gain the safety layer and the enforcement layer without doubling the hardware on the windshield.
05
Integrated Onboard Control Unit
Everything that used to need its own box on the bus now runs in one.
One unit for ADAS, DMS, BSD and AVM
Perception from every camera and sensor converges in a single SOHANIoT control unit. The side sensor's warning and the top-down view belong to the same event — not to two systems arguing about it.
An open application layer for agency apps
The unit hosts passenger-information, announcement and driver-messaging applications from the agency or approved third parties. Apps update without touching the safety core, and the agency is not locked to our roadmap for every feature.
Data out over open protocols
The unit exchanges data with back-office systems over ITxPT-aligned and standard open interfaces. Point-to-point integrations — one box for fares, another for scheduling — collapse into a single data conversation.
Remote care keeps buses on the road
Over-the-air updates, remote diagnostics and central configuration reduce depot visits. Most faults are understood from a desk; the ones that need hands-on work arrive with the diagnosis attached.
06
Operations and Dispatch
A timetable is a plan. Dispatch is what happens when the plan meets traffic.
The live picture of the fleet
Controllers see every bus on a map with on-time status, position and route. Gaps and bunching become visible while they are forming — not after passengers complain.
Events reach the right desk with context
Safety events push to dispatch with location, vehicle and video attached. The supervisor acts on the specific bus — calling the driver, adjusting the following service — without pulling reports first.
Duty records handled at the depot
Drivers sign on by face or card at shift start, and duty time, hours and vehicle assignment update automatically.
Street data flows back to planning
Actual running times, dwell times and passenger loads return to the planning team after every trip. Schedules adjust from measured performance, and each change is measured on the next round.
The Cost of Fragmented Onboard Systems
Transit agencies rarely buy one system. They buy several — a safety box, a fare link, a scheduling terminal, an annunciator, a camera system — and then pay integration costs for the rest of the contract.
- Each extra box adds its own power draw, cabling, antenna, login and firmware update cycle to maintain.
- Each point-to-point integration is a small project that must be re-tested every time either side updates.
- Each vendor boundary is a support handoff — and the data silo in between is where reporting and planning lose fidelity.
SOHANIoT's alternative is simpler: perception, applications and data out of one open unit. The agency integrates once, and extends without replacing.
The Optimization Loop
The difference between a transit network that improves and one that drifts is whether the plan and the street talk to each other. This stack closes that loop:
- Every trip produces the raw material of a better network: passenger counts, dwell times, running times and on-time performance.
- Planning tools read that data and propose changes — headways, stop timing, vehicle allocation.
- Dispatch applies the changes, and the next day's data measures the result.
The loop closes on the street, not in a spreadsheet. That is what turns a safety-and-telematics installation into a network-planning asset.
What the Open Interface Unlocks
Open data is only valuable when someone can build on it. These are the integrations the interface makes practical:
- A transit-tech vendor connects its real-time passenger app to the bus position feed, without a dedicated hardware terminal.
- The agency's data team pulls dwell-time and load records into its own analytics environment, alongside fare and AVL data.
- A third-party maintenance application reads fault codes from the unit and files them into the workshop's existing work-order system.
- A regional mobility platform aggregates on-time and load data across several operators, standardizing the feed once instead of per fleet.
Each one is a documented API call set, not a custom engineering project.
Why Agencies Choose SOHANIoT
Built and tested by the manufacturer.
Onboard units ship from our factory with certified industrial-grade hardware as standard, engineered for the vibration, temperature and electrical conditions of daily bus service.
Open by design.
ITxPT-aligned data exchange, documented APIs and an application layer for third-party software reduce fragmentation — and the integration cost that comes with it.
Right-sized rollouts.
Start with driver safety on a pilot fleet, then add enforcement, ridership analytics or collision mitigation as budgets allow. Deployment is measured in weeks, and new modules do not require replacing hardware.
Talk to an Engineer
Want the full product lineup behind this solution? Talk to a SOHANIoT engineer.
Talk to an engineerExplore the hardware that powers it — see the SOHANIoT AI Mobile DVR and ADAS-DMS Cameras.
