Hendrik-Jörn Günther

The potential of collective perception in vehicular ad-hoc networks

Most of todays Advanced Driver Assistance Systems (ADASs) rely on the data of local perception sensors. With the introduction of Vehicle-2-X (V2X) communication, the perception range of next generation vehicles will be extended even further. As a technology relying on the presence of other communication partners (network effect), any measure of increasing the market penetration rate of the technology should be pursued. This paper introduces the concept of collective perception, which aims at publishing the objects perceived by local perception sensors in addition to the envisioned self-announcement procedure of V2X enabled vehicles. The potential of the concept is studied by employing an holistic Vehicular Ad-hoc Network (VANET) simulation with a modular extension for local perception sensors. In the simulation scenarios, V2X enabled vehicles publish themselves along with their locally perceived objects. The findings unveil a significant leverage of the technology, especially in the case of low market penetration rates.

Artery: Extending Veins for VANET applications

Development and testing of novel Advanced Driver Assistance Systems (ADASs) based on Vehicle-2-X (V2X) communication is often supported by simulations. However, simulation models for Vehicular Ad Hoc Networks (VANETs) can easily become very complex due to the diverse set of involved components such as vehicle mobility, radio propagation, network protocols and application behavior. Therefore, we present an extension to the V2X simulation framework “Vehicles in Network Simulation” (Veins) with a clear separation of concerns regarding network and application aspects. Because of distinctively modeled Facilities and Application layers, this approach enables the concurrent combination of multiple VANET application sets as well as the evaluation of their interdependencies. The proposed extension to Veins, named Artery, incorporates an implementation of the ETSI ITS-G5 protocol stack following the European specifications for V2X communication. State and perception data required for ADAS algorithms are collected and provided in an extensible manner for each vehicle. Thus, Artery represents a more holistic approach in testing novel VANET applications in a network simulation environment.

Platooning at traffic lights

Congested roads not only lead to inattentive and inpatient drivers, but also have a significant negative economic impact. Especially during rush hour traffic at traffic light regulated intersections, inefficient cycle plans and the increased demand raise the need for mechanisms that increase the traffic efficiency. This paper presents a rule based algorithm for realizing platooning at traffic lights, in order to increase the throughput and to reduce the shock wave effect commonly observed at intersections. The algorithm makes use of Vehicle-2-X (V2X) communication technologies, by actively exchanging messages between the vehicles in order to set up the platoon. As a distinction to related work, the algorithm focuses on mixed traffic, in which not all vehicles are necessarily equipped with V2X technologies. The algorithm is implemented into an holistic simulation framework in order to study the effects of the parameters of the algorithm on the traffic efficiency.

Realizing collective perception in a vehicle

The introduction of Vehicle-to-X (V2X) communication enhances the perception range of a vehicle significantly. The technology, however, is subjected to the network effect: a minimum number of communication partners need to be within range, in order for V2X applications to work. Taking current vehicle replacement rates on the market into account, the time of market introduction is critical, as early adopters are likely not to profit from the technology, whilst having to account for the development costs. This contribution picks up the concept of collective perception, in which vehicles share their information gathered by their local perception sensors about objects in their vicinity. This work extends our earlier research by introducing a new message format, the Environmental Perception Message (EPM) for exchanging sensor information. Additionally, a high-level data fusion framework is presented, which separates the fusion process of local sensor data and V2X information, along with the required coordinate transformations for representing another vehicles sensor information in the recipients reference frame. We also realized the introduced framework and message in two automated vehicles and provide a performance analysis in an obstacle avoidance scenario for these vehicles on a race track. Since the vehicles share their sensor data with each other, the resulting available reaction time for avoiding the obstacle on the track increases significantly.

Collective perception and decentralized congestion control in vehicular ad-hoc networks

In conjunction with automated vehicles, Inter-Vehicle Communication represents the next ‘big thing’ towards the vision of cooperative driving, where road participants share information about their planned behaviour and where hazardous situations are solved or avoided jointly. Several requirements need to be fulfilled to reach this long-term goal. One requirement is a common information base, where all road participants are fully aware of each other. The idea of collective perception contributes to this common information base by sharing local sensor data with other road participants. Whereas most related work focuses on the aspects of sensor data fusion, we focus on the implications of an ETSI ITS G5 based network for collective perception. We present and analyse different message formats and dissemination variants for sharing sensor data. Their usability is validated in an extensive microscopic simulation study. In particular, implications caused by Decentralized Congestion Control as proposed by standardisation are assessed in a constrained environment with hundreds of vehicles.

The Effect of Decentralized Congestion Control on Collective Perception in Dense Traffic Scenarios

Abstract In conjunction with automated vehicles, Inter-Vehicle Communication represents the next ‘big thing’ towards the vision of cooperative driving, where road participants share information about their planned behaviour and where hazardous situations are solved or avoided jointly. One prerequisite to fulfill this long-term goal is a common information base, where all road participants are fully aware of each other. The idea of collective perception contributes to this information base by sharing local sensor data with other road users. Whereas most related work focuses on the aspects of sensor data fusion, we focus on the implications of an ETSI ITS G5 based network for collective perception. We present and analyse different message formats and dissemination variants for sharing sensor data. Their usability is validated in two extensive microscopic simulation studies with different traffic densities and up to several hundred concurrent vehicles. In particular, implications caused by standardised Decentralized Congestion Control are assessed in a controlled environment.