Bug Report: Index error when setting `--ego-num` to 2 or higher

[JoFrc]

Description

First of all, thank you for your great contribution to this project.

I encountered an issue when trying to run the simulation with more than one ego vehicle. The program runs correctly with the default setting of --ego-num=1. However, when I set --ego-num to 2, an IndexError occurs.

Steps to Reproduce

1. Run the simulation with the command line argument --ego-num 2.
2. The program will terminate and throw an IndexError.

Error Analysis

I've done some debugging and traced the error to the following call stack:

1. The error originates in the _load_and_run_scenario method of the LeaderboardEvaluator class, specifically at the line:
   self.manager.run_scenario()

2. This then calls the _tick_scenario method in the ScenarioManager class, where the error occurs at:
   ego_action = self._agent()

3. The execution then proceeds to the run_step method of the PnP_Agent class, calling:
   control_all = self.infer.get_action_from_list_inter(...)

4. Finally, inside the get_action_from_list_inter method, the self.generate_planning_input() function is called. The root of the problem seems to be within this function. The occ_map has a shape of (1, 1, 192, 96). When the loop for agent_i in range(self.ego_vehicles_num): reaches agent_i=1, the following line fails with an index out of bounds error:

   occ_others = self.perception_memory_bank[t]["occ_map"][agent_i]

   This is because the first dimension of occ_map is 1, but the code is trying to access the element at index 1.

Question

This behavior seems strange. Is this due to a misconfiguration on my part, or is it a bug in the current implementation for handling multiple ego vehicles?

Any help or clarification on this issue would be greatly appreciated.

[JoFrc]

--agent simulation/leaderboard/team_code/pnp_agent_e2e.py

and

--agent-config simulation/leaderboard/team_code/agent_config/pnp_config_codriving_8_10.yaml

[zsan2023]

--agent simulation/leaderboard/team_code/pnp_agent_e2e.py

and

--agent-config simulation/leaderboard/team_code/agent_config/pnp_config_codriving_8_10.yaml

I'm encountering the same issue. Have you found a solution?