US Hobbyist Uses AI and Arduino to Get 1994 Jeep Cherokee to Pass California Emissions Test

On November 13 local time, an American AI enthusiast, Mohammed Zulali, used AI tools and electronic wiring to get a 32-year-old 1994 Jeep Cherokee to pass California's emissions test. The vehicle had previously failed California's emissions test multiple times due to excessive nitrogen oxide emissions at certain speeds. Initially, the vehicle failed the 40 km/h test. After Zulali adjusted the oxygen ratio in the air-fuel mixture, it failed both the 40 km/h and 80 km/h tests. Each retest for emissions costs nearly $100 (about 668.5 yuan). The hardware Zulali used included a microcontroller priced at $3 (about 20.1 yuan), with all electronic components totaling no more than $15 (about 100.3 yuan). He used an Arduino Nano development board to connect the vehicle's engine computer to a laptop. The Cherokee uses Chrysler's SCI serial communication bus, with a communication rate of about 7,800 baud and electrical signal polarity opposite to that of a standard serial port. The Arduino Nano board was used to handle the signal inversion issue. Zulali connected the board to a breadboard, with three wires connected to the Cherokee's 6-pin diagnostic connector. The system initially could record coolant temperature and battery voltage and plot their curves. Due to mistakenly referencing data from other vehicle models, the system twice misjudged the engine air-fuel mixture as too rich, determining that fuel injection was 22% higher than it should be. In subsequent operations, the AI tool Claude was responsible for guidance and collecting data during test drives, recording engine parameters and fault codes when the vehicle was traveling at 40 km/h and 80 km/h respectively, while also gaining permission to control parameters such as fuel pressure; the AI tool Codex was responsible for writing code. Because the Cherokee's engine computer does not support reflashing firmware, Zulali had Codex write custom code to a nearby EEPROM storage chip and run it directly. Zulali said he was only responsible for wiring and driving, and did not ask a repair technician for assistance throughout the process, nor did he use Chrysler dealership-specific diagnostic equipment. Zulali has uploaded the project used to diagnose and solve the Cherokee's emissions problem to GitHub. The project is named xj-scanner, uses the GPLv3 open-source license, and is free to use.