The rapid rise of global interest in the field of autonomous driving is ushering in a new era of automobiles. With many vehicles already offering autonomous preventative safety systems, the addition of improved road infrastructure could increase the reliability and maturity of autonomous driving functions, ultimately increasing the driver's sense of safety.

Overview of the sensors for autonomous driving.
Illustration of the lateral vehicle control.

Mitsubishi Electric is developing Diamond Safety, a technology that contributes to Level 3 autonomous vehicles. Level 3 vehicles are conditionally automated driving systems that perform all driving tasks, but rely on the driver to assume control in certain situations. For example, during highway driving, drivers can take their eyes off the road and let the vehicle take over the necessary driving functions; however, drivers still need to be able to resume control after a few seconds of warning. To verify the suitability of preventative safety technology, Mitsubishi Electric uses MicroAutoBox II, a rapid control prototyping (RCP) system from dSPACE.

Satellite-Assisted Positioning Level 3 autonomous driving cannot be achieved with vehicle systems alone. The Japanese quasi-zenith satellite positioning can be employed to obtain high levels of precision and safety. The Quasi-Zenith Satellite System (QZSS, or “MICHIBIKI”) uses multiple satellites that have the same orbital period as geostationary satellites with some orbital inclinations (their orbits are known as quasi-zenith orbits). These satellites are placed in multiple orbital planes so that one satellite always appears near the zenith above the region of Japan. The system makes it possible to provide a highly accurate satellite positioning service covering close to 100% of Japan, including urban canyon and mountain terrain. The QZSS transmits its specific augmentation signals (centimeter-level augmentation information) in addition to the current positioning signals. The first satellite was launched in 2010, and by 2018, the system will feature three quasi-zenith satellites and one geosynchronous satellite. By 2023, the system will operate seven satellites to enable continuous positioning. Data received from satellite positioning systems provides the vehicle with a high level of accuracy and reliability for functions such as lane-keeping and lane-changing.

The vehicle control system, implemented completely on the Micro-AutoBox II, analyzes a multitude of sensor and vehicle data (Figure 1), including data from autonomous driving, and infrastructure components such as the forward camera, millimeter-wave radar, high-precision GNSS (global navigation satellite system) receiver, and a high-precision map. For example, the function of the lane-keeping system makes highly reliable autonomous driving possible by correlating the target driving path based on the white-line recognition with the forward monitoring camera and the target driving path obtained from a high-precision map and high-precision positioning. In addition to vehicle speed, this data is augmented by other vehicle information, which is an input for executing autonomous features and operating the actuators.

The lateral control of the vehicle utilizes a forward-looking model, as illustrated in Figure 2. The forward position of the vehicle at a future point in time and the target driving path are compared and expressed by the lateral deviation Yd and the angle θ. Then, steering control is executed by using Yd, θ, and additional vehicle information. Vehicle control in the longitudinal direction is executed by the system through speed and brake controls. Speed control is performed by comparing the speed limit data provided by the map; brake control, except for emergency braking, is set for a smooth deceleration. Stops, such as at an intersection, are executed on the basis of map data.

The dSPACE development tools allowed for a quick implementation of control algorithms developed with MATLAB® and Simulink®, making it possible to test new solutions immediately and adjust control parameters online. This is especially useful as systems gain in complexity. The tools can connect various sensors required to verify functions and alleviate communication delays. These abilities considerably reduce the number of items to be verified, thus improving productivity and liberating resources that would otherwise be required for analysis in different departments. If a system has the RCP equipment installed, the experiment software dSPACE Control-Desk allows the control and monitoring of various parameters on one screen, e.g., visually switching vehicle controls on and off, and monitoring the input/output values of various interfaces and computational values. Additionally, with its multiple interfaces, MicroAutoBox II allows for easy vehicle operation and real-time monitoring of data from measurement devices. As a result, any evaluation can easily be completed.

For more information, Click Here .