ASPICE Implementation for ML Based Software Development

Dr. Raviprakash R Salagame and Serge Lambermont

Enterprise process implementation including processes such as ASPICE  in the automotive industry is a challenging task, particularly if the organization is a small agile startup building a deep technology product. Stealth-mode organizations which are working furiously to bring their products to market often struggle with conflict between winning customers based on technology and falling short of demonstrating robustness in product development demanded by the customers. 

Autobotik worked with one such deep technology startup organization from its inception and helped them to establish the necessary infrastructure required to achieve a good process maturity level, a testimony to Autobotik’s deep expertise in understanding of automotive technology, business priorities and process knowledge. The startup is building a software product based on deep learning for the autonomous industry. Adoption of automotive processes such as ASPICE for ML based software development using agile methodology is a significant challenge and a detailed study of every activity and the tools used systematically was a major part of this project. 

Autobotik began by laying out the framework for the enterprise operating system (EOS) and key processes to support EOS. Automotive SPICE (ASPICE) was integrated into the EOS keeping in mind the big picture of business success. Several ASPICE steps such as unit testing and integration testing had to be contextualized to adopt them for ML based software development. Autobotik also assisted in setting up an exclusive productization group including hiring competent certified quality professionals to take this effort forward and eventually achieve the required formal certification. The foundation laid out by Autobotik paved the way for ASPICE Level 2 certification milestone.