Shulou(Shulou.com)05/31 Report--

This article introduces the relevant knowledge of "how to plan the module of Apollo". In the operation of actual cases, many people will encounter such a dilemma, so let the editor lead you to learn how to deal with these situations. I hope you can read it carefully and be able to achieve something!

Brief introduction

In previous versions, including Apollo 3.0, the same configuration and parameters are used to plan different driving scenarios. Although this method is linear and easy to implement, it is not flexible or oriented to specific scenario optimization. With the maturity of the Apollo system, it is more necessary to develop a more modular, scene-specific and global approach for use in different road conditions and driving cases. In this method, each driving use case is treated as a different driving scenario. This design approach is useful because in the current way, fixing a problem in a particular scenario does not affect the situation in other scenarios as it did in previous versions. In previous versions, all driving cases were considered a driving scenario, and fixing one problem affected other driving cases.

Driving scene

In Apollo 3.5, we will focus on the following three main driving scenarios:

As shown in the following figure, the lane following scene is the default driving scene, including but not limited to driving in a single lane (such as cruising) or changing lanes, following basic traffic rules or simple turns.

▲ Lane follow scene-default driving scene

In this scenario, if there are stationary vehicles or obstacles in the current vehicle lane, and the vehicle cannot safely pass through the current lane without a collision, the planning module performs the following steps:

Check to see if there are any vehicles approaching in the adjacent lane.

If safe, do a side turn from the current lane to the next lane.

Once safely through the obstacle, quickly return to the original driveway.

▲ side driving

Stop sign

For stop signs, there are two different driving scenarios:

Unprotected: in this case, the car wants to pass through an intersection with a two-way stop sign, as shown in the following figure. Therefore, the autopilot system must drive the vehicle slowly through the intersection and measure the traffic flow at the intersection before continuing.

▲ unprotected driving scene

Protected: in this case, the car wants to pass through an intersection with a four-way stop sign, as shown below. Therefore, before continuing, our autopilot system must measure the flow of traffic from the car to the stop sign and understand the position of the vehicle in the queue.

▲ protected driving scene

In order to safely pass through the stop sign area, both protected and unprotected scenarios perform the following steps:

Approaching stop sign: sense all vehicles or obstacles that are currently waiting for other stop signs.

Full stop: monitor and see if vehicles previously parked in other stop sign lanes have begun to move. It is important to make sure that all the vehicles that have arrived before have left.

Move forward slowly (crawl): check to see if any other vehicles are moving or are parked in an unprotected scene, and check to see if there are oncoming vehicles in the lane next to the current lane.

Cross the crossroads safely.

Note: the R & D team is working hard to add other driving scenarios to the planning module, one of which is dealing with traffic lights.

The architecture of the planning module in Apollo 3.5 has changed, which can reflect the modular design approach for different driving scenarios. As shown in the following figure, in the planner, each independent driving scene corresponds to the corresponding scene processor. Each driving scene has its own set of driving parameters specific to the scene, making it safer, more efficient, easier to customize and debug, and more flexible. Because each phase is divided into multiple tasks, each phase is also configurable, and each task can be deleted or created by editing the scene's configuration file config.

Architecture Diagram of the Planning Module in ▲ Apollo 3.5

Some key features:

Apollo FSM (Apollo finite State Machine): it is a finite state machine that determines the state of a vehicle according to its current location, routing scheme and high-precision map.

Planning dispenser: call the appropriate planner based on the current state of the vehicle and some other relevant information.

Planner: obtain the required context data and other information, determine the intention of the vehicle, perform the planning tasks required for that intention, and generate the planning trajectory. It will also update the context to serve future tasks.

Decision maker and optimizer: a set of stateless libraries that implement decision tasks and various optimizations. The optimizer specializes in optimizing the trajectory and speed of vehicles. Decision makers are rule-based decision makers who give advice on when to change lanes, when to stop, when to drive slowly, or when creep is complete.

Yellow boxes: these boxes are mainly reserved for future driving scenarios and developers to contribute their own driving scenarios based on real-world driving use cases.

This is the end of the content of "how to Plan the Apollo Module". Thank you for your reading. If you want to know more about the industry, you can follow the website, the editor will output more high-quality practical articles for you!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.