Current time is smaller than the first trajectory point produced by the Planning module.

[LIIHWF]

System information

• OS Platform and Distribution (e.g., Linux Ubuntu 18.04): 18.04
• Apollo installed from (source or binary): source
• Apollo version (3.5, 5.0, 5.5, 6.0): 7.0

Steps to reproduce the issue:

I'm using MODE_MOCK to simulate Apollo Autopilot, and found that the first trajectory point produced by the Planning module can be larger than the current time. Here is how I triggered this problem:

• Update perception information for every 0.05 seconds.
• The Planning module will be triggered to execute once per 0.05 seconds.
• However, only the first produced trajectory contains a point with relative_time=0. For subsequent trajectories, the relative_time of the first point is 0.1 seconds, which exceeds the current time (0.05 seconds later).

Is that the correct result? Do I need to configure any parameter to improve it?

[YuqiHuai]

Maybe the issue you described is causing SimControl to produce weird scenarios ...

https://user-images.githubusercontent.com/34195403/228130073-e2a86225-22c9-401f-a4b5-9c9e8abf84e2.mp4

[LIIHWF]

@YuqiHuai Thanks for your video!

So the update frequency of the Planning module must be below 10Hz? Is the time interval of the trajectory points determined by any configuration parameter?

Does this issue also affect real-world control?

[YuqiHuai]

@LIIHWF Based on https://github.com/ApolloAuto/apollo/blob/b00c030ac02c54f98579d3d3e879dbf52f0d1d53/modules/dreamview/backend/sim_control_manager/dynamic_model/perfect_control/sim_perfect_control.h#L167

I would guess the developers are expecting the Prediction frequency to be 10 Hz, and since Planning is triggered by Prediction then Planning should also be producing messages at 10 Hz. I haven't looked into the details of trajectory points' time intervals and I am also hoping someone more knowledgeable about this can determine what is causing the issue and why SimControl is shown to be teleporting back and forth.

I also guess this wouldn't have a real-world effect. Because Apollo sends control commands to LG SVL and I haven't seen anything weird about ego car movement in SVL-based scenarios besides the shaking issue that might have been caused by delays.

[starrye]

@LIIHWF Based on

https://github.com/ApolloAuto/apollo/blob/b00c030ac02c54f98579d3d3e879dbf52f0d1d53/modules/dreamview/backend/sim_control_manager/dynamic_model/perfect_control/sim_perfect_control.h#L167

I would guess the developers are expecting the Prediction frequency to be 10 Hz, and since Planning is triggered by Prediction then Planning should also be producing messages at 10 Hz. I haven't looked into the details of trajectory points' time intervals and I am also hoping someone more knowledgeable about this can determine what is causing the issue and why SimControl is shown to be teleporting back and forth.

I also guess this wouldn't have a real-world effect. Because Apollo sends control commands to LG SVL and I haven't seen anything weird about ego car movement in SVL-based scenarios besides the shaking issue that might have been caused by delays.

apollo的planning日志有什么异常的输出吗？比如一直replan什么的

[LIIHWF]

apollo的planning日志有什么异常的输出吗？比如一直replan什么的

Yes. In my case, the planning module reports warnings in the log:

[trajectory_stitcher.cc:156] current time smaller than the previous trajectory's first time

The parameter is_replan in the output is always true.

[YuqiHuai]

@starrye @LIIHWF Same here. I also noticed planning keeps sending "rerouting requests".