Brussels View: Driverless Merging Ahead - Inside GNSS - Global Navigation Satellite Systems Engineering, Policy, and Design

Brussels View: Driverless Merging Ahead

Researchers in the PRoPART project developed a 10-centimeter accurate and robust position determination system for autonomous vehicles, demonstrating a safe merge of a fully automated heavy truck between two cars driving at speed in an adjacent lane.

A European cooperative project focusing on autonomous vehicles and advanced driver assistance systems, PRoPART has developed an autonomous road transport system fully enabled to make use of the new Galileo Open Service Navigation Message Authentication (OS-NMA), the Galileo E5 Alternative Binary Offset Carrier (AltBOC) signal and multi-frequency measurements for absolute positioning and time information. The project utilized vehicle-mounted radar and cameras in combination with a robust GNSS module, supplemented by radar-equipped reference stations along the route. The combined sensor and GNSS data enabled high-reliability positioning in the ten-centimeter range.

PRoPART project coordinator Stefan Nord from Research Institutes of Sweden (RISE) said the most outstanding technical innovation developed by the project is its system for co-operative perception, supported by precise position and timing. “The main challenge was to perfectly synchronize vehicle sensor systems, roadside units and the GNSS signals,” he said.

“When vehicles and road infrastructure share information with one another,” Nord said, “the time of validity of the information is critical. When this information relates to perception data, the ‘pose’—position, orientation, speed, estimated error—of the perception sensors in a common reference frame is also essential for interpreting the data. Any error in estimation will feed through to inaccuracies in the system perception and potentially reduce efficacy or safety, particularly if the system does not have high integrity.”

Key GNSS Contribution

Project partner Fraunhofer Institute for Integrated Circuits (IIS) provided the GNSS receiver for the PRoPART system. “We used our receiver development platform GOOSE—GNSS receiver with open software interface,” said Matthias Overbeck, Fraunhofer IIS Group Manager for Precise GNSS Receivers. “We put the whole thing together, from antenna, front end, digital baseband processing, to position solution. GOOSE is the only available development platform for GNSS receivers and it can provide an own hardware solution independent from commercial receivers.”

The platform, Overbeck said, is dedicated to software developers, mobile communication operators, and system developers. “GOOSE is available for integration in commercial PC or as an embedded platform for different applications, e.g. integration in vehicles or together with a smart antenna.”

“GOOSE is already able to perform vector tracking and deep coupling in real-time to give a robust tracking solution,” said Overbeck. “This was one requirement of the PRoPART project, to be able to bridge signal interruptions for short time periods, like in tunnels or under bridges.

“We are also able to exploit the new Galileo services, one of the first GNSS receivers that is ready for the soon-to-be-launched OS-NMA. Additionally, the intermediate frequency signals can be recorded, processed and replayed with the platform. This allows repeatable real-time tests and makes GOOSE essentially a portable GNSS laboratory.”

GOOSE also provides a 10MHz reference input and output and has a pulse per second (PPS) output (1 Hz—max. 25Hz) as well as an Ethernet connection and two full-speed USB ports to enable open and flexible access.

For PRoPART, the GOOSE platform hosts two separate positioning engines. Fraunhofer’ s Deeply Coupled Tracker (DCT) fused data from an IMU (Inertial Measurement Unit). “The DCT collocates the information with the GNSS antenna signals in a vector tracking loop implementation,” said Overbeck.

The European Collaborative Process

PRoPART was funded by the European GNSS Agency (GSA), under the European Union’s Horizon 2020 research budget. For the EU, efficient transport is a fundamental priority in its larger vision of a smarter, greener and more inclusive society. The Union is also promoting an ambitious ‘zero-deaths, zero injuries’ target for road transport.

“The main idea behind our project,” said Nord, was to develop and enhance an RTK software solution by exploiting the distinguishing features of Galileo and combining it with other positioning and sensor technologies for greater precision, safety and efficiency. Furthermore, we introduced and developed several novel Day-2 C-ITS services [V2X] and demonstrated new vehicular communication functions for automated driving. We have a double-difference Kalman filter that fuses all measurements for consumption by the rest of the system, and also feedback to enable rapid integer ambiguity resolving, and the DCT, which implements a deeply integrated inertial/code phase tracking solution for generating a navigation solution.”

Day-2 C-ITS refers to an innovation phase defined by the Cooperative Intelligent Transport Systems (C-ITS), a platform. Day 1 services focus on exchanging information enhancing foresighted driving. Day 2 services improve the service quality and share perception and awareness information. Day 3+ adds further sophisticated services like sharing intentions, supporting negotiation and cooperation that paves the way towards cooperative accident-free automated driving.

For now, this kind of groundbreaking work remains inevitably collaborative, because single companies, even big vehicle manufacturers, don’t yet have sufficient in-house know-how in all the necessary specialized technology areas. Thus, RISE first set out to assemble a team. “We made contact with Fraunhofer IIS, because we knew about their GOOSE platform and their knowledge in GNSS receiver capabilities and antenna design,” said Nord. “But the project needed to bring in other positioning technologies as well.”

Ceit-IK4 is the Spanish research institute that developed PRoPART’s UWB solution as a complimentary system to the GNSS positioning module. This helped to guarantee a continuous, accurate and reliable position in situations where GNSS signals are not available, for example in tunnels, thick forests or deep urban canyons.

Hungarian company Commsignia was responsible for system communications, prototyping two new features, a real-time correction message (RTCM) and collective perception message (CPM) for free space and object distribution. Commsignia V2X modules, located in the roadside units, served as communication hubs, receiving data from all of the external data collection equipment. On the test track, three identical roadside units covered 900 meters of road surface in total, providing full RTCM, CPM and UWB coverage.

Another key partner, Swedish company Waysure, provided the GNSS RTK software that combined GNSS measurements with in-vehicle motion sensors, odometry and IMU data, and UWB and reference station radar observations from the roadside units. Germany’s Baselabs, were responsible for the so-called ‘environment module’, which generated a unified representation of the environment around the vehicle, fusing all data from the radars and smart cameras on the truck and the radars mounted on the roadside units. The company also provided the crucial ‘decision module’ that gave the green light for the automated lane change, based on the information generated by the environment module.

The self-driving truck, no small piece of the puzzle, was made by Scania, and finally, project coordinator RISE itself provided system architecture expertise and, via their subsidiary AstaZero, the test track in Sweden where PRoPART was demonstrated.

“The UWB ranging technology has the potential to provide positioning where GNSS cannot be used, e.g. in tunnels,” Nord said. “That means the combination of these two technologies can provide for indoor and outdoor applications. Relying only on one technology is not sufficient when developing systems that should cover all challenges for autonomous driving. For now, we need to continue working on more efficient UWB-antennas, to guarantee a higher signal stability.”

Nord also said he thinks UWB should be a standard component for proximity ranging and location. “The potential is confirmed by the efforts made in both automotive and mobile telephone industries,” he said.

Remaining obstacles to full implementation of PRoPART-like applications on real roads include laws and regulations that still need to be pushed forward. “The testing and certification of these vehicles imposes challenges on what methods and tools should be used. In particular, for Sweden, both Trafikverket, the Swedish Transport Administration, and Transportstyrelsen, the Swedish Transport Agency, are engaging in autonomous trials and investigating the impact on policy, society, business models, and infrastructure, and allowing limited autonomous operation on public roads.”

As to whether the public is ready to accept autonomous vehicles, Nord said, “In well-known areas, for example open mines, ports and logistics centers, with lower-speed requirements for vehicles than on public roads, acceptance of fully autonomous vehicles is higher. Autonomous driving is likely to roll out in these controlled and understood environments earlier than in public spaces in Europe.”

Now that PRoPART has been demonstrated successfully, market exploitation falls back to the individual partners. “The main point of the project was to enable the partners to develop competitive subsystems that can be included in a variety of solutions,” said Nord. “The different contributions of course are owned by the different partners. In particular, the industrial partners Waysure, Commsignia and Baselabs will continue to develop their respective products. Their main customers are vehicle OEMs as well as TIER 1s for the in-vehicle components, and road infrastructure owners for the roadside units. Demonstrating the performance of a system and its components in projects such as PRoPART can help to convince these customers of the readiness and value of precise positioning and ITS.”

Although PRoPART is a European project, the technologies demonstrated in the project are not bound to specific countries or regions. “These could be used in other parts of the world,” concluded Nord, “although different regions have very different considerations when it comes to standards, regulations and what is required to engage and do business with both public and private companies.”