bron Automotiv:
COMPONENTS
Draft standard on HD-maps for AVs issued
Draft standard on HD-maps for AVs issued
By Roma Nazarov
20th May 2021
The ITS market has entered the phase of explosive growth throughout Europe.
In Russia, more than 10 regions are now deploying elements of smart infrastructure. While positive news for progress in AVs, it raises specialists’ concerns over disunity and poor compatibility of regional and national systems. The industry is facing questions, like, would an autonomous truck rolling along the E8 aurora borealis transport corridor and pass smoothly the Finnish-Russian border or stop motionless?
“Flawless data exchange between regional ITSs is critically important,” thinks Sultan Zhankaziev, head of the chair of traffic safety and management at Moscow Automobile and Road Construction Institute (MADI). To fix that issue, MADI has now revealed draft standards on architecture and data precision of digital road maps used for administering autonomous traffic. Due for approval early this year, the standards will find practical implementation in the on-going works on creating HD-maps of 8,000km (4,970 miles) of Russian federal motorways. In 2023, it will step down to regional roads.
“The package of standards on digital road maps will find implementation in all infrastructural projects for preparation of safe use of autonomous vehicles on public roads,” said Vasily Kurguzov, head of ITS department at Russian road research institute Rosdornii, the federal head institution for the development of ITS.
Vital role
Maps matter much more to self-driving vehicles than they do to human drivers. All the co-operation between CAVs and the governing ITS is done via digital maps. The layers on maps for AVs contain high-precision data for localization and maneuvering. Maps for CAVs contain also dynamic layers with real-time traffic lights and neighboring vehicles’ locations and speeds and more.
There are many driving scenarios with situations when assistance from ITS is indispensable. Maneuvers such as pulling out at a T-junction with no clear view of the main road would be impossible without it. A number of other scenarios, although feasible for pure autonomous systems, can be enacted with greater safety and comfort thanks to external data. In one instance, it helps to pass an obstacle on a highway with one lane in each direction without slowing down even if view on the on-coming traffic is poor.
In a real-world experiment, MADI team set an AV on a short route in urban canyons involving some of these situations. The researchers found that a vehicle driven by only machine vision required five enforced interventions by a safety driver in just one attempt. When artificial vision was combined with satellite navigation, there were only two intervention instances. However, there was a drastic improvement to one intervention per ten attempts when the vehicle was supported by an ITS featuring road-side cameras in the challenging locations.
In another series of virtual and physical tests, the researchers found that platoons of non-connected vehicles consisting of more than five vehicles were prone to a chain reaction of the vehicles in the tail of the platoon on acceleration and braking of the vehicles in the front, leading to uncomfortably sharp braking.
On the cross-way
Debate is still going on among professionals over architecture and precision and types of data. Enthusiasts of autonomous driving insist on less-is-more, generally, because they are skeptical about the speed of deployment of ITS. However, the downsides of this approach range from a shorter window for decision-making to higher demands on on-board computing power.
For that reasons, the vast majority of AV developers embrace use of maps while disagreeing over their architecture. Some prefer to process point clouds while others supply the vehicle with just corridor margins or even drive along a virtual rail to reduce volumes of on-board calculation.
“The proposed architecture allows all that approaches,” Andrey Vorobyev, deputy head of MADI’s ITS competence center, said ahead of the public hearing over the draft standards. “We aim only to provide conditions for vehicles to continue obtaining data when switching between regional ITSs.”
Vorobyev said that the new standards are supposed to be of a recommendations nature, allowing different mapping concepts to co-exist in the market, trusting to evolution. As for unification of local standards in neighboring countries, it is urgently needed, he thinks. For that purpose, Russian regulators mean to propose the local use case as a base for future international standards on digital maps for CAV-ITS co-operation.