AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Revisiting OOD Generalization in Programmatic RL

Amirhossein Rajabpour, Kiarash Aghakasiri, Sandra Zilles, Levi Lelis

ICML 2026 · Reinforcement learning › Visual and representation RL · approach: Learned state and action representations

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use a cautious intrinsic reward (beta 0.5 instead of 1.0) for the neural policy (ablation)DDPG policy in TORCS · TORCS OOD tracksgeneralization to unseen tracks: CR 1608M (all crashed) -> 1:48 (0.76) on G-TRACK-2Improved
lower beta in the reward, which slows the car (ablation)DDPG policy in TORCS · G-TRACK-1 training tracklap time: 54 -> 1:17Hurt
use cautious reward when training on AALBORG (ablation)DDPG policy in TORCS · ALPINE-2 and RUUDSKOGENfraction of seeds generalizing: CR 1688M -> 3:13 (1.00); CR 3232M -> 2:46 (1.00)Improved
augment observation with the agent's last action in a fully connected network (instead of LSTM or ConvNet)PPO in partially observable KAREL · KAREL TOPOFF 100x100average return: 0.15 -> 1.00 (vs LSTM)Improved
use a_{t-1} with partial observation instead of ConvNet with full observationPPO in KAREL · KAREL STAIRCLIMBER 100x100average return: 0.00 -> 1.00Improved
use a_{t-1} instead of ConvNet on HARVESTERPPO in KAREL · KAREL HARVESTERaverage return: 0.90 -> 0.59 (small); 0.00 -> 0.04 (100x100)Mixed
use an LSTM for partial observability instead of a_{t-1}PPO in KAREL · KAREL STAIRCLIMBER smallaverage return: 1.00 -> 0.13Hurt
neural policy (DQN) vs programmatic PSM on a tight-gap testPARKING · PARKING testsuccess rate: 0.16 (PSM) vs 0.18 (DQN)No clear effect
compare generalization gap between train and testPSM vs DQN in PARKING · PARKINGsuccess rate drop: PSM 0.26 -> 0.16; DQN 0.86 -> 0.18Mixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy