Experimental results acquisition issue

[gaohan-cmd]

Hello author, I would like to inquire about how I can obtain the accuracy data for Table 2: Comparison of different LLMs and LMMs on the task of embodied task planning, and Table 3: The average execution success rate of generated action steps for different RGB image collection strategies in scene perception, which were mentioned in the paper. I couldn't find any information in the code regarding the generation of this data. Can you answer this question for me? Your help would be greatly appreciated!

[Gary3410]

Hi, thank you for your interest in this work!
Our current TaPA can only generate high-level planning and cannot be executed directly in the simulator. So the results in Table 2 and Table 3 are as stated in the paper, with three volunteers recruited for scoring.
As for how the navigation data in Table 3 are generated, you can take a look at this section. We employed different strategies to capture the visual information, and after that, we gave them to TaPA to generate the planning and asked for volunteers to score them