- Automated Driving
- Automated Driving Systems
- Vehicle to Vehicle Communications
- Advanced Sensors
Vehicle-to-Everything Communications Spurred by Automated Driving
I’ve been a long-time proponent of the idea that automated vehicles (AVs) should all be equipped with vehicle-to-everything (V2X) communications. Vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and other pathways can provide important real-time information that complements what the on-vehicle sensors of an AV can see.
Ultimately, an AV needs to safely navigate through its environment solely based on what it can see and hear, just as human drivers do. However, like human drivers, AVs don’t yet have Superman’s X-ray vision. The sensors can only detect what is visible within the spectrum of that sensor. Cameras, radar, lidar, or thermal imagers cannot see around the corner of a building, through the truck in front, or detect the pedestrian behind the delivery van about to cross the street.
Communicating to Build Understanding
V2V communications can also be a valuable mechanism for AVs to signal intent about what the vehicle is about to do. In 2016, a Waymo AV struck the side of a bus because it misjudged what the human bus driver was going to do approaching a work zone. V2V may have helped avoid that.
Most automated driving system (ADS) developers in China have been using cellular-V2X (C-V2X) for some time now, including Baidu and AutoX. US-based ADS companies are now expanding use of V2X and remote sensing. Ford and its ADS partner Argo AI recently began deploying sensor pods above challenging intersections in Miami, Florida. The pods are equipped with cameras, radar, and lidar that get a bird’s-eye view of vehicles and other road users in the area of the intersection. A C-V2X radio is used to communicate with approaching AVs when there is a hazard that the vehicle should be aware of.
In Las Vegas, Nevada, Motional and its predecessor companies Delphi and Aptiv have used V2I roadside units to get traffic signal phase and timing information to correlate with what sensors are seeing for several years. Motional has partnered with Derq, a developer of traffic monitoring systems. Similar to Ford’s Miami project, Derq has installed cameras above several Las Vegas intersections that are tied to computers running robotic perception software.
Observing and Interacting
Initially, the data from Derq will be used offline by Motional as an input signal to simulations and to verify ground truth information about the scenarios the AVs encounter. Later, information will be transmitted to approaching AVs in real time to provide a beyond line-of-sight view. With earlier awareness of hazards such as pedestrians or cyclists that may not be visible, the AV can respond sooner, providing a more comfortable ride for passengers by avoiding sudden braking. The system should also benefit safety by allowing AVs to avoid higher risk scenarios.
Derq will use the basic and personal safety message formats defined in the Society of Automotive Engineers J2735 standard. Now that the US Federal Communications Commission has ruled on the use of C-V2X for automotive safety, automakers are expected to start deploying the technology as soon as 2022. Those human-driven vehicles will also be able to benefit from this enhanced situational awareness from hazard alerts. While an AV needs to be able to safely navigate through its environment based only on what its on-board sensors can detect, adding additional layers of data to the mix should enhance comfort and safety in the more complex scenarios in dense, urban environments.