Automated Driving with Cooperative Perception Using Millimeter-Wave V2V Communications for Safe Overtaking

Abstract

The combination of onboard sensors on vehicles with wireless communication has great advantages over the conventional driving systems in terms of safety and reliability. This technique is often called cooperative perception. Cooperative perception is expected to compensate for blind spots in dynamic maps, which are caused by obstacles. Few blind spots in dynamic maps can improve the safety and reliability of driving thanks to the additional information beyond the sensing of the onboard sensors. In this paper, we analyzed the required sensor data rate to be exchanged for the cooperative perception in order to enable a new level of safe and reliable automated driving in overtaking scenario. The required sensor data rate was calculated by the combination of recognition and vehicle movement to adopt realistic assumptions. In the end of the paper, we compared the required sensor data rate with the outage data rate realized by the conventional V2V communication and millimeter-wave communication. The results showed the indispensability of millimeter-wave communications in automated driving systems.

Automated Driving with Cooperative Perception Based on CVFH and Millimeter-Wave V2I Communications for Safe and Efficient Passing through Intersections

“…Classifying by road shape, a two-lane roads are where the most traffic accidents take place in Japan. Therefore, in the previous work, this type of traffic accident is analyzed for safe automated driving [ 8 ]. Unsignalized intersections are where the second-largest amount of accidents occurs in Japan—accounting for 25%.…”

“…In general, there are two ways to recognize an object, i.e., specific object recognition and general object recognition. In [ 8 ], specific object recognition using edge points was performed to recognize a vehicle. Since the recognition target is only a vehicle in this analysis, which is the same as in the above works, a specific object recognition was adopted.…”

“…The received power is formulated as follows: where is the transmission power, and are the antenna gains for the direct and reflected wave, and are the optical path length for direct and reflected waves, is the absorption factor at 60 GHz by oxygen as 15 dB/km, c is the speed of light, is a carrier frequency, k is , and is the complex reflection coefficient. A basic analysis of the effect of vibration on fading and height diversity has already been performed in a previous work [ 8 ]. Therefore, we focus on the effect of height diversity in the case of V2I communications.…”

“…This algorithm consists of a sharing sensor data part and recognition part. The sharing sensor data part is based on the previous work [ 8 ] and we checked that it works in the real environment in [ 18 ]. The recognition part is based on the typical model base recognition process of CVFH, which was also released in the robot operating system and used for object recognition in practice.…”

“…In this paper, sharing raw sensor data is assumed for distributed verification, and the utilization of millimeter-wave communications is considered to tackle the problem of communication resources and to benefit from the advantages of sharing raw sensor data. This paper is the extended work of our previous work and the contribution of this paper consists of two aspects as follows [ 8 , 9 ]: The first aspect is the derivation of the required sensor data rate for passing through an intersection safely. Although millimeter-wave communications are believed to play an important role in sharing raw sensor data, the minimum required amount of raw sensor data that is transmitted for safe automated driving via millimeter-wave communications is unclear.…”