Autonomous Process Assessment Framework for ASPICE Capability Optimization in Software-Defined Vehicles

Abstract

The automotive industry is undergoing a significant transformation toward Software-Defined Vehicles (SDVs), where software functionality increasingly determines vehicle performance, safety, connectivity, and user experience. As software complexity continues to grow, organizations face mounting challenges in maintaining process capability, ensuring compliance with Automotive SPICE (ASPICE), and achieving predictable development outcomes. Traditional ASPICE assessments provide valuable insights into process maturity but are inherently retrospective, relying on periodic evidence collection and assessor-driven evaluations. Consequently, capability weaknesses are often identified only after they have already impacted project performance, resulting in increased rework, delayed releases, and elevated quality risks. This paper proposes a Dynamic ASPICE Capability Prediction Framework (DACPF) that leverages Artificial Intelligence (AI), engineering analytics, and process mining techniques to forecast future ASPICE capability achievement within Software-Defined Vehicle development programs. The framework continuously collects engineering evidence from development ecosystems including requirements management, source code repositories, continuous integration pipelines, testing environments, and quality management systems. Machine learning models analyze engineering indicators such as traceability completeness, review effectiveness, defect leakage, test coverage, and process conformance to generate predictive capability assessments and maturity forecasts. A conceptual case study involving a large-scale SDV development program demonstrates how the proposed framework can improve capability visibility, reduce assessment preparation effort, and provide early warning indicators for capability risks. Results indicate that predictive capability monitoring can significantly enhance organizational readiness for ASPICE assessments while supporting continuous improvement initiatives. The proposed framework represents a shift from reactive compliance management toward predictive process governance, enabling organizations to proactively optimize process maturity and sustain capability achievement in increasingly complex automotive software environments.