Skip to main content
— Presentation at ITST 2015 —
December 22, 2015
The 14th International Conference on ITS Telecommunications (ITST2015) was held in Copenhagen, Denmark on 2nd to 4th December 2015. In this conference, researchers from universities and enterprises in various parts of the world (especially in Europe) discussed technologies on Vehicle to X (V2X) communication and automated driving.
Research and Development Group, Hitachi, Ltd., made a presentation regarding the method for data representation and interface of surrounding information, recognized by on-board sensors and/or digital map, for automated driving, with the title "Extended Electronic Horizon for Automated Driving".
Fig. 1 Hierarchical hybrid data representation of
surrounding environment information
One of representative related works is ADASIS (Advanced Driver Assistance System Interface Specification), a de-facto standard data representation and interface specification for ADAS (Advanced Driver Assistance System). In the ADASIS specification, position relationships of surrounding information are represented along the roads in the driving route (e.g. there is an intersection 30 m ahead along the road), and therefore it is suited for the development of longitudinal driving control applications such as Adaptive Cruise Control (ACC). However, it would not be suited for automated driving requiring lateral driving control, because it does not support lane-level and detailed data representation (e.g. which lane other vehicles are located, how the shapes of lanes are).
In the presentation, we proposed a hierarchical and hybrid data representation method enabling efficient and flexible data provision of lane-level and detailed surrounding information, required for automated driving (Fig. 1). There are two main features in the proposed method. Firstly, it organizes two-layered data representation architecture of road-level abstracted layer (Layer 1) and lane-level detailed layer (Layer 2). We adopt the ADASIS specification (version 2) for Layer 1 as it is, and extend the data representation into lane-level for automated driving in Layer 2. Thanks to this approach, it becomes possible to support automated driving with backward compatibility to existing products based on ADASIS specification. Secondly, the proposed method represents position relationships with surrounding information by two types of coordinates, path-based one like ADASIS specification (e.g. there is an vehicle 30m ahead of the same lane as the ego vehicle) and spatial one (i.e. relative position vector from the ego vehicle). The former is fit for quick macroscopic recognition of surrounding information, while the latter allows detailed microscopic recognition. This makes it possible to flexibly develop various driving control applications including automated driving, by adaptively using either of or both coordinates according to their purpose.