Similar scenarios in one log file

[QuanyiLi]

Hi,

I really appreciate your effort in building this amazing dataset! Every map detail and surrounding object are well recorded and of high quality. One minor question is that when I play the scenarios from one log, why there are a lot of similar scenarios? What is the relationship between .db log and scenarios? My config for loading log is as follows:

 "DATASET_PARAMS": [
                'scenario_builder=nuplan_mini', # use nuplan mini database (2.5h of 8 autolabeled logs in Las Vegas)
                'scenario_filter=one_continuous_log',  # simulate only one log
                "scenario_filter.log_names=['2021.05.12.22.00.38_veh-35_01008_01518']",
                'scenario_filter.limit_total_scenarios=1000',  # use 1000 total scenarios
            ],

[gianmarco-motional]

So one log represent one run of data collection with the car. A scenario is a portion of the log (the .db file) extracted and preprocessed with some given parameters (duration, past/future, distance between scenarios etc).

Therefore scenarios built from the same log (especially if they are close in time) will be similar. If you want diversity I would advise not using one_continuous_log and enabling shuffling in the scenario_filter parameters, or you can use just a different pre-set configuration file.