Working prototype (simulated, no physical hardware)

Edge AI EV Optimizer

On-device load balancing for EV charging, simulated end to end in Wokwi.

Project from the Emertxe IoT internship (common brief across the cohort). The implementation and contribution below are mine.

Emertxe IoT online internship:

Overview

An ESP32-based EV charging station optimizer that predicts charging demand and supports load balancing using on-device inference, MQTT telemetry, and a ThingsBoard dashboard. Simulated end to end in Wokwi.

Problem Statement

Predicting charging demand and balancing load across a charging station.

Proposed Solution

ESP32 firmware reads voltage and current inputs and runs a lightweight AI model on the device to support load-balancing decisions, with telemetry over MQTT to a ThingsBoard dashboard with alerts and remote control. Simulated end to end in Wokwi.

System Architecture

Edge AI EV Optimizer system architecture. System flow: Simulated sensors → ESP32 (on-device inference + decision logic) → MQTT → ThingsBoard dashboard (alerts, remote control).Simulated sensorsESP32 (on-device inference + decision logic)MQTTThingsBoard dashboard (alerts, remote control)

Technologies Used

  • ESP32
  • Embedded C
  • MQTT
  • ThingsBoard
  • Wokwi

My Contribution

My work covers the ESP32 firmware, on-device inference and decision logic, MQTT telemetry, and ThingsBoard integration for the simulated end-to-end flow.

Development Status

Working prototype

Future Scope

  • Concept: explore it as the load-management layer for the SmartPayPlug charging station concept (not built).