# Safety through feedback in Constrained RL

arXiv: https://arxiv.org/abs/2406.19626 · HTML: https://mlsanity.com/paper/2406.19626/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 23 of its 39 references that are in the map.
- First posted on arXiv 28 Jun 2024.

## Builds on (most cited first, 23 of 23)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Generative Adversarial Imitation Learning](https://mlsanity.com/paper/1606.03476/) (2016)
- [Constrained Policy Optimization](https://mlsanity.com/paper/1705.10528/) (2017)
- [Unifying Count-Based Exploration and Intrinsic Motivation](https://mlsanity.com/paper/1606.01868/) (2016)
- [#Exploration: A Study of Count-Based Exploration for Deep Reinforcement Learning](https://mlsanity.com/paper/1611.04717/) (2016)
- [PEBBLE: Feedback-Efficient Interactive Reinforcement Learning via Relabeling Experience and Unsupervised Pre-training](https://mlsanity.com/paper/2106.05091/) (2021)
- [End-to-End Safe Reinforcement Learning through Barrier Functions for Safety-Critical Continuous Control Tasks](https://mlsanity.com/paper/1903.08792/) (2019)
- [Large-Scale Study of Curiosity-Driven Learning](https://mlsanity.com/paper/1808.04355/) (2018)
- [IQ-Learn: Inverse soft-Q Learning for Imitation](https://mlsanity.com/paper/2106.12142/) (2021)
- [Safety-Gymnasium: A Unified Safe Reinforcement Learning Benchmark](https://mlsanity.com/paper/2310.12567/) (2023)
- [Safe Reinforcement Learning in Constrained Markov Decision Processes](https://mlsanity.com/paper/2008.06626/) (2020)
- [Trial without Error: Towards Safe Reinforcement Learning via Human Intervention](https://mlsanity.com/paper/1707.05173/) (2017)
- [Reward Uncertainty for Exploration in Preference-based Reinforcement Learning](https://mlsanity.com/paper/2205.12401/) (2022)
- [Constrained Update Projection Approach to Safe Policy Optimization](https://mlsanity.com/paper/2209.07089/) (2022)
- [Inverse Constrained Reinforcement Learning](https://mlsanity.com/paper/2011.09999/) (2020)
- [EX2: Exploration with Exemplar Models for Deep Reinforcement Learning](https://mlsanity.com/paper/1703.01260/) (2017)
- [Safe Reinforcement Learning with Linear Function Approximation](https://mlsanity.com/paper/2106.06239/) (2021)
- [Benchmarking Constraint Inference in Inverse Reinforcement Learning](https://mlsanity.com/paper/2206.09670/) (2022)
- [Interactively Learning Preference Constraints in Linear Bandits](https://mlsanity.com/paper/2206.05255/) (2022)
- [Provable Safe Reinforcement Learning with Binary Feedback](https://mlsanity.com/paper/2210.14492/) (2022)
- [Imitate the Good and Avoid the Bad: An Incremental Approach to Safe Reinforcement Learning](https://mlsanity.com/paper/2312.10385/) (2023)
- [Long-term Safe Reinforcement Learning with Binary Feedback](https://mlsanity.com/paper/2401.03786/) (2024)

## Built on by (most cited first, 5 of 5)

- [A Survey of Safe Reinforcement Learning and Constrained MDPs: A Technical Survey on Single-Agent and Multi-Agent Safety](https://mlsanity.com/paper/2505.17342/) (2025)
- [MoralityGym: A Benchmark for Evaluating Hierarchical Moral Alignment in Sequential Decision-Making Agents](https://mlsanity.com/paper/2602.13372/) (2026)
- [TraCeS: Learning Per-Timestep Constraint-Violation Credit from Sparse Trajectory-Level Labels](https://mlsanity.com/paper/2504.12557/) (2025)
- [Safe Reinforcement Learning with Preference-based Constraint Inference](https://mlsanity.com/paper/2603.23565/) (2026)
- [Redistribution-based Cost Inference Improves Sparse Safe Offline RL](https://mlsanity.com/paper/2608.12306/) (2026)

## Abstract (arXiv)

In safety-critical RL settings, the inclusion of an additional cost function is often favoured over the arduous task of modifying the reward function to ensure the agent's safe behaviour. However, designing or evaluating such a cost function can be prohibitively expensive. For instance, in the domain of self-driving, designing a cost function that encompasses all unsafe behaviours (e.g. aggressive lane changes) is inherently complex. In such scenarios, the cost function can be learned from feedback collected offline in between training rounds. This feedback can be system generated or elicited from a human observing the training process. Previous approaches have not been able to scale to complex environments and are constrained to receiving feedback at the state level which can be expensive to collect. To this end, we introduce an approach that scales to more complex domains and extends to beyond state-level feedback, thus, reducing the burden on the evaluator. Inferring the cost function in such settings poses challenges, particularly in assigning credit to individual states based on trajectory-level feedback. To address this, we propose a surrogate objective that transforms the problem into a state-level supervised classification task with noisy labels, which can be solved efficiently. Additionally, it is often infeasible to collect feedback on every trajectory generated by the agent, hence, two fundamental questions arise: (1) Which trajectories should be presented to the human? and (2) How many trajectories are necessary for effective learning? To address these questions, we introduce textitnovelty-based sampling that selectively involves the evaluator only when the the agent encounters a textitnovel trajectory. We showcase the efficiency of our method through experimentation on several benchmark Safety Gymnasium environments and realistic self-driving scenarios.