Motion2D - Pure Planning and Policy Learning Experiments

[mz1231]

Create BiRRT planning approach (motion2d_birrt_approach.py) to run on Motion2D and generate results.

Test llm_ppl_approach.py on Motion2D and generate results.

Here are the Pure Planning Results (Evaluates BiRRT on Motion2d across 5 seeds and 5 passage counts (0, 1, 3, 5, 7))

Here are the Pure Policy Learning Results (Synthesizes 5 policies (one per approach seed) and evaluates each on the same 5 passage instances)