Cooperative driving in the context of an interaction between the driver and a semi-autonomous vehicle is not the focus of this article. Instead, how a level 4 or a level 5 autonomous vehicle should cooperate with any road users or systems that could influence the decision of self-driving cars is the main focus of this article. This subject is a new research topic in the field of Intelligent Transportation Systems (ITS).

We need self-driving cars with the ability to drive cooperatively because of many reasons such as:

• Recent researches and experiences show that not all driving scenarios can be solved by an autonomous vehicle alone. These vehicles need to have more information to make more accurate and faster decisions in critical situations.
• Due to the high complexity of level 5 autonomy development, cooperative driving can be used as a bridge to achieve safe level 4 vehicles.
• An autonomous vehicle is part of a traffic management system. Optimizing the behavior of such a vehicle that is an element of this system does not mean optimizing the entire transportation system.

Self-driving car as a source of data

The data collected by each autonomous vehicle can be beneficial for other cars or traffic management systems to regulate and optimize overall traffic quality. The amount of data collected by an autonomous vehicle using all the sensors available in the vehicle is huge, and usually, the row data is not stored. Only the useful part of the data for decision-making is extracted for the vehicle decision, and the whole data is deleted after the decision is made.

This information can also be helpful for other road users driving in the same area or useful for statistical applications. There is another piece of data collected by the sensors because the sensors have a wide field of view. Sometimes, this data does not directly affect the decision-making process of the ego vehicle but can help another car or the entire traffic system avoid accidents. This piece of information is often deleted or ignored, and unfortunately, this is actually missing information that can still be useful.

Sef-driving car as an actuator

In many cases, we have to use an autonomous vehicle as an actuator such as the following:

• Receive a request from an autonomous vehicle stating that an uncertain situation has occurred. The autonomous vehicle cannot resolve the issue alone, for example, if police had stopped the autonomous vehicle.
• Improving traffic quality requires central management to overview the entire controlled area, not just from a vehicle perspective.

An autonomous vehicle is designed to make decisions in any simple or complex scenario. There are still driving conditions such as intersections, which can be very complicated for self-driving cars, depending on the number of vehicles at the intersection or the size of the intersection, or the speed limit.

Using an autonomous vehicle as an actuator is crucial in terms of safety. Ultimately, the vehicle is responsible for safe driving and decision-making, not the remote operator or infrastructure that submitted the request. Any connection of an autonomous vehicle to another vehicle, or infrastructure, or operator control center is a source of error propagation. An error in one of these devices outside the vehicle can result in an incorrect decision by the vehicle.

Mixed traffic scenarios

The infrastructure that allows mixed traffic for autonomous vehicles and human-driven vehicles must be highly intelligent and provide services for autonomous applications and road safety. Safe and secure communication is essential for an autonomous vehicle and very useful for a human-driven car.

It is the task of an autonomous vehicle or human in human-driven vehicle to ensure road safety for all road participants. Safe and secure communication is more useful for increasing the availability of autonomy. More communication and information exchange between subsystems in a system can reduce road safety. Communications can potentially increase the likelihood of system failure (vehicle, cloud, roadside unit, road infrastructure, or vehicle control center). This means, more communication between elements of a system requires more safety and security mechanisms in an autonomous vehicle and infrastructure as well to increase the availability and safety of autonomous functionalities at the same time.

A new standard for cooperativeness

SAE International has published a new standard, SAE J3216 for Cooperative Driving Automation (CDA), to achieve consensus among experts and a guideline for implementing cooperativeness for self-driving cars. This new SAE standard J3216 intends to improve traffic and safety in autonomous driving.

The J3216 is based on four classes with increasing degrees of cooperation:

1. Status sharing
2. Intent sharing
3. Agreement seeking
4. Prescriptive

The J3216 addresses the operational and tactical part of the dynamic driving task and does not deal with the strategic part.

Summing up,

Many applications such as teleoperated driving, infrastructure-controlled automated driving, or cooperative adaptive cruise control are research and development. These applications are switching between centralized and distributed control of the vehicle. We need to pay more attention to programs that require high scheduling and timing issues when transferring responsibility from autonomous control to infrastructure or operator control or in another direction.

We still have to solve the level 3 automation problem and transferring responsibility from the vehicle to the driver and vice versa in implementing cooperative driving.

Realizing mixed traffic is a fundamental requirement for achieving autonomy in the automotive industry and requires intelligent, reliable, safe, and secure infrastructure.We have to redefine many things and define new concepts.

We need to ask ourselves whether autonomous driving technology can tackle current types of roads or topologies or whether we need to change them to increase road safety or reduce the cost of developing autonomy.

Source: medium.com