Self-Learning Audit Orchestration Model for Continuous ASPICE Process Governance
DOI:
https://doi.org/10.63282/3050-922X.IJERET-V4I4P120Keywords:
ASPICE, SQA Audits, Process Governance, Continuous Compliance, Automotive Software Engineering, AI, Audit Automation, Audit Orchestration, Machine Learning, Reinforcement LearningAbstract
Automotive SPICE (ASPICE) compliance is essential for ensuring quality and safety in automotive software development. Traditional ASPICE assessments are periodic, resource-intensive, and reactive, often leading to compliance drift between official audits. This paper introduces a novel Self-Learning Audit Orchestration Model (SLAOM) for continuous process governance. By leveraging reinforcement learning and natural language processing (NLP), SLAOM continuously ingests artifacts from engineering lifecycles, maps them to ASPICE Process Reference Models (PRM), and orchestrates automated mini-audits. The system self-learns from auditor feedback to minimize false positives, providing a real-time compliance posture. Experimental validation demonstrates an 84% reduction in manual audit preparation time and a 91% accuracy rate in predicting Capability Level compliance.
References
[1] VDA QMC, Automotive SPICE Process Reference Model/Process Assessment Model v4.0, 2023.
[2] C. Jones, "Continuous Process Verification in Safety-Critical Systems," IEEE Transactions on Software Engineering, vol. 48, no. 3, pp. 245-258, March 2022.
[3] ISO/IEC 33020:2019, Information technology — Process assessment — Process capability assessment model, ISO, 2019.
[4] ISO/IEC 33020:2015 – Process Measurement Framework
[5] ISO/IEC 15504 (SPICE) Standard.
[6] ISO 26262-1:2018 Functional Safety Standard.
