Honda Research Institute on data management system for ADAS and AD systems

Automotive IQ sat down with Antonello Ceravola, Principal Scientist at Honda Research Center Europe, and talked about data management system for ADAS and AD testing at Honda RC, ahead of the conference where Mr. Ceravola also will be presenting.

What is your role at Honda when it comes to ADAS and AD topics?

My focus is on software architecture, software design, integration infrastructures and tools.

I am in charge on the software design of ADAS and AD projects and on the research and development of middleware/tools for supporting creation of such systems.

(We are organized with a matrix structure between competences and projects)

One of the main challenges in the area of testing of ADAS and autonomous cars is the management of test data. What, in your experience, are the main challenges when it comes to test data?

The main challenges we experienced are:

- Data formats: Handling the plurality of data formats coming out from the different projects/platforms

- Data Management Infrastructure: Having a data management infrastructure able to support big-data, different recording formats, automatic/manual labelling, efficient search and fast access to relevant recording segments for the different projects (we are currently developing such a data management infrastructure, as well as looking around on the coming infrastructure in the market)

- Teaching: Instructing and educating teams for a correct data creation and data infrastructure usage

- Global Coordination: coordinating activities in this domain at global level, since, the data management problem is present at Honda world-wide level.

What solutions are there currently available in the industry? 

Few novel data management systems, suitable for the ADAS/AD domain are coming in the market, and a long standing consortium (ASAM ODS), which has been created for small-size data and now moving to the big-data requirements coming from the ADAS/AD domain. 

What is Honda doing in this area? 

We are both data producer (in our work on intelligent systems on different platforms: robots, car, ….) and researching/developing infrastructures able to manage such data for the development and test of such complex intelligent systems.

What is the main improvement to the existing data management systems that yours brought to the table?

We are designing a streamlined data management process, implemented in our data management infrastructure. Together with this we are designing advanced user interface able to give to researchers all the flexibility necessary to have a quick access to the big amount of recordings we are constantly producing. Moreover we are participating in standardization committees in order to contribute to the community and make sure our effort can be used in the long run. 

What will be the contents and structure of your presentation at the IQPC event? 

My presentation will be a case study, showing our approach and lesson learned in the research and development of data management infrastructures for ADAS/AD systems and more.