Are self-driving cars safe? As the automotive industry moves towards higher levels of automation, it’s important for the answer to this question to always be yes. At TomTom, it’s our vision to create a safe, connected and autonomous world – and a big role in making autonomous driving safer is played by ADAS and HD maps.
Maps – ADAS and HD – are one of the four pillars of autonomous driving. Together with onboard sensors, driving policy and actuators, they form the technology that enables automated and autonomous driving. HD maps specifically improve localization to centimeter-level accuracy and sensor perception, which leads to safer path planning by automated driving systems.
Accelerated by the spread of COVID-19, fully autonomous driving and robot-taxis seem further away on the horizon, compared to expectations from past years on the pace of progress in automation. However, industry consensus remains that society would benefit from reaching the highest levels of vehicle automation in many ways. For example, it would offer a solution for increased demand of autonomous package delivery services as well as personal transportation over public transport.
For safe autonomous driving to become a reality, you need detailed HD maps that are constantly updated.
In this blog post, we will take a look at the fundamental principles of HD maps for autonomous driving, why they are so important and how they contribute to the safety of autonomous driving.
An important requirement for high-quality HD maps is for them to be complete. This includes map attribution reflecting reality with centimeter-level accuracy, but also extends to the complementary services around it. Complete HD maps with a rich feature set can only be achieved through a multi-sourcing approach of survey vehicles, community input and crowdsourced car sensor data.
In addition to HD maps, a portfolio of complementary services is needed. Services like TomTom AutoStream make sure that detailed maps are constantly updated through secure and data-efficient map delivery.
We also need to consider the different needs of lower levels of automated driving. In this case, we would need a less granular ADAS map that has larger coverage. This is perfectly suited for lower-level automation features such as adaptive cruise control (ACC).
TomTom was the first to launch HD maps in 2015, and many other companies have followed suit since then. However, map-making is a unique skill that requires decades of experience before it can be deployed on a global scale. In a future with millions of automated driving cars, we need a map that can process all that data and still ensure all these have access to the latest HD and ADAS maps.
With our industry-leading transactional map-making platform, TomTom is best positioned to enable safe autonomous driving with ADAS and HD Maps on a global scale. This brings us to the third principle of HD maps to combat safety issues concerning self-driving cars.
It’s important for HD maps to be constantly updated. Being able to reflect changes in reality on the map as soon as possible from the moment they happen is critical for accuracy and, ultimately, safety in a self-driving car.
Let’s take the example of a new traffic sign. Its existence should be reflected in the map as quickly as possible. This can be quantified as reality to map time (R2M). The shorter the R2M, the more accurate the HD map and therefore the safer it is for AD systems to rely on.
At TomTom, we strive to minimize R2M to maximize safety, as we have been doing for years with the industry-leading real-time TomTom Traffic service. With TomTom Traffic, we process hundreds of millions GPS data points to create an accurate representation of traffic density on the road, with cycles times of less than a minute.
For self-driving cars to be safe, HD maps should be complete, proven and fast. At TomTom, we are relentless in perfecting our HD and ADAS mapping technology. Stay tuned as we continue to announce exciting news and partnerships in 2020 and beyond.