Verify the mechanism behind the degradation at 128 LiDAR bins

Determine whether the increased mean lateral deviation observed with the 128-bin LiDAR representation is caused by increased network-input capacity and the unchanged training budget rather than by the representation's information content.

Background

The 128-bin representation produces higher mean lateral deviation than the 32- and 64-bin representations, despite providing finer angular resolution. The paper interprets this pattern as potentially reflecting increased input capacity without additional task-relevant obstacle information, combined with the same training budget. That explanation remains speculative because the experiments do not isolate capacity, optimization time, or information resolution.

References

A reading consistent with the rest of the table is capacity rather than information---128 bins enlarge the observation by $64$ dimensions without supplying obstacle detail this task appears to need, and the policy is given the same $500{,}000$-step budget in which to learn the larger input---but this remains a hypothesis consistent with the ordering rather than a demonstrated mechanism.

— CORAL: Curriculum-Optimized Reward Adaptation for LiDAR-Based Goal-Directed Urban Driving  (2608.14332 - Saleem et al., 14 Aug 2026) in Section 4.4, 'Ablation: The LiDAR Observation'