Scale-CDA: A Scalable Aftermarket Platform to Democratize Cooperative Driving Automation in Production Cars
Scaling cooperative driving automation (CDA) to production passenger vehicles requires an affordable retrofit platform that can accommodate heterogeneous OEM Controller Area Network (CAN) signals and advanced driver-assistance system (ADAS) commands. Scale-CDA addresses this challenge by building on OpenDBC and openpilot, which provide vehicle interfaces and Level-2 automation support for more than 300 car models. The proposed open-hardware and open-software prototype integrates commodity edge computing, camera sensing, Wi-Fi, cellular communication, and a CAN adapter for less than $1,000. It exchanges telemetry and cooperative messages through MQTT over Wi-Fi~6 or LTE, avoiding costly DSRC or C-V2X sidelink radios. In moving-vehicle experiments, Wi-Fi~6 achieved a mean round-trip time of 5.25~ms and mean negotiated physical-layer rates of 98.51~Mb/s for transmission and 109.17~Mb/s for reception, with approximately 2% of observations exceeding 50~ms. These results demonstrate feasibility for non-safety-critical CDA applications. Scale-CDA also introduces a Generative AI interface that combines camera observations, CAN data, and connectivity messages through the Model Context Protocol. The model generates semantic message intents and structured MetaActions, while deterministic adapters validate and encode cooperative messages and map admissible actions to existing Level-2 functions without allowing direct actuator control. An on-road construction-zone demonstration validates driver advisory, message generation, and speed-related MetaAction pathways. Scale-CDA provides a reproducible platform for connectivity and GenAI research on supported production vehicles.