The previous installment in this series detailed how operational design domains are pre-defined to determine where an autonomous vehicle can operate, with the function of the vehicle limited to within its ODD. This installment explores how to handle situations when the vehicle encounters conditions beyond the scope of its designated ODD.
Imagine you’re about to head out in your car. You might choose to delay your departure if it's foggy or, following snowfall, wait until the snow melts later in the day. These are sensible decisions. However, it’s not always feasible to cancel plans based on the mere possibility of fog or heavy rain.
While driving, if fog sets in, you’re likely to reduce your speed, and if visibility worsens, turn on your hazard lights and carefully pull over to the shoulder. If you’re lucky enough to be near a shopping center when heavy rain hits, you might stop there for a break. These judgment calls are necessary not only for bad weather but also in situations like a flat tire or car trouble.
An autonomous vehicle undergoes rain testing.
Autonomous vehicles also need to make judgments and respond to situations like these. In Level 3 autonomous driving, a human is always in the driver's seat and can take over when the system cannot cope. If the vehicle determines that it can’t continue operating autonomously, it prompts the driver to take control using audible or visual alerts on the dashboard.
This is known as a take-over request, and when it’s issued by the autonomous vehicle, the driver must promptly take control. So, in Level 3, human drivers take control in the event that the autonomous driving system cannot handle the situation.
Level 4 doesn't require a human driver, so the vehicle must handle all situations on its own. The simplest response is for the vehicle to stop if it encounters a problem. However, even stopping involves a choice between braking sharply to stop quickly or gradually slowing down. If there’s imminent danger, an emergency stop might be necessary, but this can carry risks such as rear-end collisions or passenger injuries, so careful judgment on how to stop is required.
Stopping on the road can obstruct traffic and increase the risk of rear-end collisions, so sometimes it’s necessary to pull over to the shoulder. This requires a minimal level of continued autonomous operation, including steering and braking. The vehicle might need to make a more sophisticated response, such as continuing to a bus stop or parking area before coming to a halt.
When it becomes difficult to safely continue autonomous driving, the vehicle performs what is called a minimal risk maneuver (MRM) to avoid danger. The goal of this MRM action is to reach what is known as a minimal risk condition (MRC).
Reaching an MRC might include decelerating sharply to stop; decelerating gradually; pulling over to the shoulder; or driving to the next parking area before stopping. However, when an issue arises with an autonomous vehicle, the choice of which MRC to aim for and how to perform the corresponding MRM can vary depending on the nature of the problem and the surrounding conditions.
In cases where a more advanced MRM, such as "driving to the next parking area," is required, more accurate obstacle detection and situational assessment are needed. Executing advanced MRMs autonomously is quite challenging.
For autonomous vehicles to become commonplace in society, the technology must be able to handle a wide range of road environments and effectively manage associated risks. Tackling these challenges individually by each manufacturer would be inefficient. Instead, a collaborative approach is needed to advance the technology.
This installment looked at how minimal risk maneuvers and minimal risk conditions are used to handle challenging autonomous driving scenarios. In the next installment, the focus shifts to Autoware, the world’s first open-source software for autonomous driving.
Toshihide Ando | TIER IV Fellow
Toshihide joined TIER IV in 2019, where he has served in roles including vice president of technology before assuming his current position as fellow. Previous experience includes software development and R&D at a major Japanese automotive parts manufacturer.
TIER IV is always on the lookout for passionate individuals to join our journey. If you share our vision of making autonomous driving accessible to all, get in touch.
Visit our careers page to view all job openings.
If you’re uncertain about which roles align best with your experience, or if the current job openings don’t quite match your preferences, register your interest here. We’ll get in touch if a role that matches your experience becomes available, and schedule an informal interview.
Inquiries
Social Media
X (Japan/Global) | LinkedIn | Facebook | Instagram | YouTube
More
.