3arXiv cs.AI (Artificial Intelligence)·
Explainable RL with saliency maps applied to Air Traffic Control routing
A new arXiv preprint applies explainability techniques to reinforcement learning agents operating in a simplified Air Traffic Control environment, where the agent learns to route flights around no-fly zones. Saliency maps are used as a preliminary explainability mechanism to identify which input features most influence the agent's decisions. The work targets the broader challenge of building trust in AI systems deployed in safety-critical domains.
Related guides (1)
Related events (8)
Physics-aware policy distillation makes deep RL agents interpretable via decision tree surrogates
A new arXiv preprint presents a policy distillation framework that transfers a high-performance TD3 reinforcement learning agent into a shallow decision tree surrogate for interpretable continuous control, demonstrated on the Inverted Pendulum benchmark. The method uses physics-aware features and 'Noisy Oracle Rollouts' to match teacher performance while providing global and local interpretability. A key finding is a fundamental trade-off: discretizing control induces Bang-Bang actuation and a bimodal limit cycle, though BIBO stability is preserved. The work targets safety-critical deployment contexts where regulatory compliance and human-agent trust require explainability.
AgenticRL: Self-refining LLM-guided reward design and policy refinement for UAV navigation
AgenticRL is a framework that uses a multimodal GPT agent to automate reward function generation, policy training via PPO, and closed-loop self-refinement for UAV navigation tasks. The agent evaluates trained policies through diagnostic feedback, identifies failure modes, and iteratively refines rewards without human intervention. Evaluated across five navigation tasks, the closed-loop refinement improves policy behavior by 71% over initial rewards, with sim-to-real transfer achieving 91% real-world success rate and 94% sim-to-real accuracy.
ACPO: Adaptive Clip Policy Optimization improves RLVR training for LLM reasoning
A new arXiv preprint provides theoretical analysis of Reinforcement Learning from Verifiable Rewards (RLVR) updates, identifying off-policy degree and gradient expectation as key factors governing update dynamics. The authors show that differences in gradient steps per rollout substantially affect importance sampling ratio distributions and which tokens dominate updates. Based on this analysis, they propose Adaptive Clip Policy Optimization (ACPO), which adjusts clipping boundaries per token group by empirical variance of importance sampling ratios, outperforming DAPO and CISPO baselines on 3B and 7B models across math, tabular QA, and logic benchmarks.
Scaling Up Reinforcement Learning for Traffic Smoothing: A 100-AV Highway Deployment
Berkeley AI Research (BAIR) deployed 100 RL-controlled autonomous vehicles into real rush-hour highway traffic on Interstate 24 near Nashville to dampen stop-and-go waves and reduce fuel consumption. The RL controllers were trained in data-driven simulations built from real highway trajectory data, using only local sensor inputs (speed, lead vehicle speed, gap) to enable decentralized deployment on standard vehicles. Reward design balanced wave smoothing, energy efficiency, safety, comfort, and adherence to human driving norms. The paper documents the sim-to-real transfer challenges encountered during this large-scale field experiment.
ExpRL: RL-based mid-training using human QA data as reward scaffolds for LLM reasoning
ExpRL proposes an automated approach to LLM mid-training that replaces manually curated reasoning traces with large corpora of human-written QA data used as reward scaffolds rather than imitation targets. Reference solutions are hidden from the policy and used only to construct problem-specific grading rubrics, enabling dense process-level rewards that reinforce partial progress and intermediate reasoning steps. On challenging math reasoning benchmarks, ExpRL outperforms SFT, sparse-reward GRPO, and self-distillation as an RL initialization strategy, with additional mixed-domain experiments suggesting broader applicability.
Emergent language in multi-agent RL proposed as generative methodology for studying AI consciousness
A new arXiv preprint proposes using emergent language (EL) in multi-agent reinforcement learning as a generative methodology for studying consciousness-relevant structure in AI systems, contrasting with existing discriminative or architectural approaches. Agents begin with minimal language exposure and develop communication under task pressure alone, aiming to avoid artifacts from human language priors. As a proof of concept, the authors show agents develop self-referential communication including an echo-mismatch detection circuit that emerges from environmental affordances rather than task structure or architecture.
LongTraceRL: Reinforcement Learning for Long-Context Reasoning via Search Agent Trajectories and Rubric Rewards
LongTraceRL is a new RL training framework for improving long-context reasoning in LLMs, addressing limitations of existing RLVR methods. It constructs challenging training data using multi-hop questions from knowledge graph random walks and tiered distractors derived from search agent trajectories (high-confusability: read but uncited; low-confusability: seen but unopened). A rubric reward provides entity-level process supervision along reasoning chains, applied only to correct responses to prevent reward hacking. Experiments across three LLMs (4B–30B parameters) on five long-context benchmarks show consistent improvements over strong baselines.
Improving Model Safety Behavior with Rule-Based Rewards
OpenAI has developed a method called Rule-Based Rewards (RBRs) that trains models to behave safely without requiring extensive human data collection. The approach uses explicit rules to generate reward signals during training, offering a more scalable alternative to traditional RLHF-based safety alignment. This represents a practical contribution to alignment methodology from a Tier 1 lab.