Stima ingests CAN bus data from each vehicle, processes it at the edge, and surfaces actionable alerts before a pack fails. Here's exactly how that works.
Five steps from raw vehicle data to a confident dispatch decision.
The Stima Edge Module (SEM-1) connects to the vehicle's OBD-II port. For commercial EVs without standard OBD-II, a CAN bus tap harness is available for Bajaj RE, TVS iQube, and Zembo Pikipiki models. Installation takes under 10 minutes per vehicle with no tools required.
SEM-1 samples voltage, current, cell temperatures, and state-of-charge at 10Hz and stores readings in a local ring buffer. When a 2G/3G/4G connection is available, the module flushes the buffer to Stima's regional data centers in Nairobi and Lagos. No reading is ever lost — the buffer holds 72 hours.
An extended Kalman filter running on Stima's servers estimates the true state of health (SoH) for each pack, accounting for temperature correction, depth-of-discharge history, and calendar aging. The model was trained on 18,000 discharge cycles across LFP and NMC chemistries common in Africa-market EVs.
A gradient-boosted classifier runs daily against each pack's rolling 30-day telemetry. Packs with declining SoH trajectories receive a degradation score between 0 and 100. Scores below 65 trigger a yellow alert; below 45 trigger red. Fleet managers see both the score and the trend line.
Each morning at 05:00 local time, Stima generates a charge routing plan for the day's shifts. The plan factors state-of-charge, predicted route distance from your existing dispatch system, and charger station wait times (updated in real time from Stima's charging network partners). The plan is delivered as a JSON feed or displayed directly in the dispatch dashboard.
What's under the hood — for operators who want the specifics.
SEM-1 speaks OBD-II (ISO 15765-4 CAN), J1939, and a proprietary BMS protocol used by three major Africa-market EV manufacturers. Data is normalized to a common schema before being written to the time-series database.
Supported chemistries: LFP, NMC, LTO. Nominal pack voltages from 24V to 96V.
The base Kalman filter model is fine-tuned per vehicle make using online learning. After 30 days of operation on a given fleet type, the model's mean absolute error in state-of-health estimation drops from ±6% to ±2.4%.
Model updates deploy nightly without service interruption.
All telemetry is encrypted in transit using TLS 1.3. At rest, data is stored with AES-256 encryption on AWS Africa (Cape Town) and AWS Asia Pacific (Singapore) instances. Each operator's data is logically isolated — no cross-tenant reads are possible at the database layer.
Data residency options: EU, Africa, Asia-Pacific.
Stima exposes a REST API with OpenAPI 3.0 documentation. Webhook support for real-time alerts. Native integrations exist for Tuma Dispatch (Kenya), FleetBoss (Nigeria), and any fleet management platform that accepts JSON webhooks.
SDK available in Python and JavaScript.
Drivers receive pre-shift battery status and their assigned charge stop via the Stima Driver app (Android, iOS). Push notifications alert drivers if state-of-charge drops below the safe threshold mid-route. Works on sub-$80 Android handsets.
Supports English, Swahili, French, Hausa.
Weekly fleet health reports are emailed automatically. The analytics module shows pack-level SoH curves, charge efficiency trends, and comparative degradation across vehicle cohorts. Exportable as PDF or XLSX.
Custom dashboards available on Growth and Enterprise plans.
We'll walk you through a live demo on real fleet data — yours or ours.
Request a Demo