The DOI matches its Crossref record; 27 of the 31 references it cites were found in a registry or an index, and 4 are not checked yet; 2 of the 6 sources its claims cite were checked for support.
Safe Reinforcement Learning via Shielding
Open accessAbstract
Reinforcement learning algorithms discover policies that maximize reward, but do not necessarily guarantee safety during learning or execution phases. We introduce a new approach to learn optimal policies while enforcing properties expressed in temporal logic. To this end, given the temporal logic specification that is to be obeyed by the learning system, we propose to synthesize a reactive system called a shield. The shield monitors the actions from the learner and corrects them only if the chosen action causes a violation of the specification. We discuss which requirements a shield must meet to preserve the convergence guarantees of the learner. Finally, we demonstrate the versatility of our approach on several challenging reinforcement learning scenarios.
Verified 2026-10-24 · 4 references not checked yet
Overview
AI snapshot
Written from the paper’s full text.
TL;DR
The paper proposes shielded reinforcement learning, where a reactive shield synthesized from a temporal-logic safety specification and an environment abstraction monitors learner actions and replaces them only when needed to avoid unsafe behavior. It proves correctness and minimal interference, gives conditions for preserving convergence, and demonstrates safety and often faster learning in four simulated domains.
Key findings
- Shields can enforce safety during both learning and execution without changing the reward signal, and they are minimally interfering: they block an action only when it could lead to a specification violation under the abstraction.
- In the water-tank example, the product game had 602 states, and the shield prevented opening the inflow valve above level 93 and forced opening when the level became too low.
- Shield synthesis was fast in the reported experiments: under 2 seconds for the grid-world cases, about 2 seconds for the self-driving car, under 1 second for the water tank, and under 7 seconds for Pacman.
- In the 15x9 grid world, only the shielded setting with a ranking of 3 actions and no penalty found the optimal path; in the water-tank experiments, shielded Q-learning and SARSA reached optimal policies in significantly fewer episodes than unshielded versions; in the self-driving car case, the shielded learner never crashed and learned faster; in Pacman, the shielded agent cleared the game from the first episode.
Methods
- The approach specifies safety in linear temporal logic, converts it to a deterministic safety automaton, builds a conservative finite-state abstraction of the MDP environment, and solves a safety game to synthesize a shield.
- The shield receives the learner’s action or ranked list of actions and executes the highest-ranked safe action; it deviates only if every ranked action is unsafe, in which case it chooses some safe action.
- The paper proves correctness and minimal interference, constructs a product MDP to argue convergence for standard MDP learners, and evaluates the method in simulations: 9x9 and 15x9 grid worlds, a 480x480-pixel self-driving car task, the water-tank controller, and an 18x5 Pacman-like maze, using tabular Q-learning, deep Q-networks, and approximate Q-learning.
Limitations
- The paper states that the main downside is the need for an approximate model or abstraction indicating when actions are unsafe, and argues this is unavoidable when allowed actions depend on the environment state.
- The convergence argument requires that when the learner’s ranking contains no safe action, the shield’s choice among safe actions follows a fixed probability distribution over time, and it assumes the learner can observe the shield state.
- Evident from the text: the evaluation is entirely simulation-based, with no real-world deployment; the approach relies on the abstraction being conservative, and the shield must remain active after learning because unsafe actions may still be part of the learned policy.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
45 Own interpretation · 10 Own result · 4 No basis given · 1 Partly supported · 1 Not supported · 1 No full text of the sources · 1 Not checked yet · accuracy not yet measured
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The results show that only the unshielded versions experience negative rewards.”
§ Safe Reinforcement Learning via Shielding > Experiments
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Fig. 11 shows that both shielded (dashed lines) and unshielded Q-learning and SARSA experiments (solid lines) do reach an optimal policy.”
§ Safe Reinforcement Learning via Shielding > Experiments
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In all of them, the learning performance of the shielded agents improved compared to the unshielded case.”
§ Safe Reinforcement Learning via Shielding > Conclusion
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“As expected, in the shielded case, pacman is always able to clear the game starting from the first episode.”
§ Safe Reinforcement Learning via Shielding > Experiments
plaindirectSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, the shielded implementations reach the optimal policy in a significantly shorter time than the unshielded implementations.”
§ Safe Reinforcement Learning via Shielding > Experiments
plaindirectSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“the computational cost of the construction of the shield depends on the complexity of the specification and a very abstract version of the system”
§ Safe Reinforcement Learning via Shielding > Safety in Formal Methods.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“since the shield corrects unsafe actions in the learning and deployment phases.”
§ Safe Reinforcement Learning via Shielding > Related Work Safety in RL.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Even though the inner working of learning algorithms is often complex, shielding with respect to critical safety specifications may be manageable (as we demonstrate in upcoming sections).”
§ Safe Reinforcement Learning via Shielding > Environment
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, to reason about when exactly a specification violation cannot be avoided, we have to give a (coarse finite-state) abstraction of the environment dynamics.”
§ Safe Reinforcement Learning via Shielding > Safety, Abstractions, and Games
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We introduce a new approach to learn optimal policies while enforcing properties expressed in temporal logic.”
§ Safe Reinforcement Learning via Shielding > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Probably not a substantial claim. 17
- It may state two separate claims as one. 17
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“An exploration process is called safe if no undesirable states are ever visited, which can only be achieved through the incorporation of external knowledge (García and Fernández 2015;Moldovan and Abbeel 2012).”
§ Safe Reinforcement Learning via Shielding > Related Work Safety in RL.
What the sources say
- Not checked — no full text of the sourcecites García and Fernández 2015
- Disputedcites Moldovan and Abbeel 2012 sourceExperimental — accuracy not yet measured
The Safety Constraint section (p. 3) defines safety as preserving ergodicity with a user-specified probability, not as never visiting an undesirable state. The Safety Counter-Examples section (p. 3) explicitly argues against defining safety by safe and unsafe states. The paper uses a prior belief and information learned during exploration, but does not say that incorporating external knowledge is the only way to achieve safety. (LLM, conf=0.82)
Checked by a reasoning model — may be wrong.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Unlike the approach in the paper, their technique does not work with function approximation, which prevents it from being used in complex scenarios.”
§ Safe Reinforcement Learning via Shielding > Safety in Formal Methods.
What the sources say
- Partly supportedcites Wen, Ehlers, and Topcu 2015 sourceExperimental — accuracy not yet measured
Section V (Examples; page unavailable) shows that learning time and iterations grow as the grid grows, and describes a trade-off between performance and computation cost. This supports a scalability concern, but the paper does not say that its technique cannot work with function approximation or that it cannot be used in complex scenarios. (LLM, conf=0.76)
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The plot in Fig. 10 shows that the accumulated rewards for unshielded RL (red, dashed) increases over time, but the car still crashes at the end of the simulation.”
§ Safe Reinforcement Learning via Shielding > Experiments
plaindirectSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In scenarios with |rank t | = 1 (red) or with penalties (solid), the agent computes a suboptimal path.”
§ Safe Reinforcement Learning via Shielding > Experiments
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Only the shielded version with |rank t | = 3 without penalty (blue, dashed) finds the optimal path, resulting in a higher average reward.”
§ Safe Reinforcement Learning via Shielding > Experiments
plaindirectSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Furthermore, the shielded version is not only safe, but also tends to learn more rapidly.”
§ Safe Reinforcement Learning via Shielding > Experiments
hedgedconditionalSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The shielded version without punishment (blue, solid) learns more rapidly than the unshielded learner and never crashes.”
§ Safe Reinforcement Learning via Shielding > Experiments
plaindirectSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
In the paper
“For specifications in LTL, it is known how to check if it is a safety language and how to compute a safety automaton that represents it (Kupferman and Vardi 2001).”
§ Safe Reinforcement Learning via Shielding > The set of all distributions on X is denoted by Distr(X).
What the sources say
- Not checked — no full text of the sourcecites Kupferman and Vardi 2001 source
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“An abstraction of an MDP describes how its executions can possibly evolve, and provides the needed information about the environment to allow planning ahead w.r.t. the safety properties of interest.”
§ Safe Reinforcement Learning via Shielding > Safety, Abstractions, and Games
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“and is independent of the state space components of the system to be controlled that are irrelevant for enforcing the safety specification.”
§ Safe Reinforcement Learning via Shielding > Safety in Formal Methods.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“It ensures safety and minimum interference.”
§ Safe Reinforcement Learning via Shielding > Environment
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The shield is computed upfront from the safety part of the given system specification and an abstraction of the agent's environment dynamics.”
§ Safe Reinforcement Learning via Shielding > Environment
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This boundary helps to separate the concerns, e.g., safety and correctness on one side and convergence and optimality on the other”
§ Safe Reinforcement Learning via Shielding > The Thirty-Second AAAI Conference on Artificial Intelligence (AAAI-18)
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The latter is necessary as the valve has to stay open for at least three time steps.So as the inflow may be up to 2 liters/second during this time and the outflow may be 0, there is otherwise an overflow risk.”
§ Safe Reinforcement Learning via Shielding > Safety, Abstractions, and Games
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This approximation may have finitely many states even if the MDP has infinitely many states and/or is only approximately known.”
§ Safe Reinforcement Learning via Shielding > Safety, Abstractions, and Games
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Last but not least, the shielding framework is compatible with mechanisms such as function approximation, employed by learning algorithms in order to improve their scalability.”
§ Safe Reinforcement Learning via Shielding > The Thirty-Second AAAI Conference on Artificial Intelligence (AAAI-18)
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The algorithms we present for the computation of shields make relatively mild assumptions on the input-output structure of the learning algorithm (rather than its inner working).”
§ Safe Reinforcement Learning via Shielding > Environment
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Reactive synthesis enforces ϕ s by solving a safety game built from ϕ s and an abstraction of the environment in which the policy is to be executed.”
§ Safe Reinforcement Learning via Shielding > Safety, Abstractions, and Games
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Planning ahead does not require the environment dynamics to be completely known in advance.”
§ Safe Reinforcement Learning via Shielding > Safety, Abstractions, and Games
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“All of the above approaches have in common that a faithful, yet precise enough, abstraction of the physical environment is required, which is not only difficult to obtain in practice, but also introduces the mentioned computa-tional burden.”
§ Safe Reinforcement Learning via Shielding > Safety in Formal Methods.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“and provides a basis for the convergence analysis of a shielded RL algorithm.”
§ Safe Reinforcement Learning via Shielding > The Thirty-Second AAAI Conference on Artificial Intelligence (AAAI-18)
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This ensures that paths that model traces that cannot occur in M are rejected by ϕ M as early as possible.”
§ Safe Reinforcement Learning via Shielding > Safety, Abstractions, and Games
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“As the shield is generated using the specification, it plans ahead for this not to happen, so it must prevent the opening of the inflow valve if the level is above 93.”
§ Safe Reinforcement Learning via Shielding > Safety, Abstractions, and Games
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our setup introduces a clear boundary between the learning agent and the shield.”
§ Safe Reinforcement Learning via Shielding > The Thirty-Second AAAI Conference on Artificial Intelligence (AAAI-18)
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Given that the environment is often represented as an MDP, such an abstraction has to be conservative w.r.t. the behavior of the real MDP.”
§ Safe Reinforcement Learning via Shielding > Safety, Abstractions, and Games
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
- Rests on an assumption: The environment is often represented as an MDP.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Control methods based on RL partly address this problem, but do not typically incorporate any correctness guarantees.”
§ Safe Reinforcement Learning via Shielding > Safety in Formal Methods.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Shielding exploits this possibility.”
§ Safe Reinforcement Learning via Shielding > Conclusion
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Since an unsafe action is always mapped to a safe one, this does not pose a problem and the agent never has to learn to avoid unsafe actions.Consequently, the shield is (again) needed during the learning and execution phases.”
§ Safe Reinforcement Learning via Shielding > Shielded Reinforcement Learning
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our experiments show, however, that in applications in which safe learning is needed, the effort to construct an abstraction is well-spent, as our approach not only makes learning safe, but also shows great promise of improving learning performance.”
§ Safe Reinforcement Learning via Shielding > Conclusion
boosteddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The agent assigns a punishment r t+1 < 0 to a 1 t and learns that selecting a 1 t at state s t is unsafe, without ever violating ϕ s .”
§ Safe Reinforcement Learning via Shielding > Shielded Reinforcement Learning
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- The authors state this as an assumption: The environment can be modeled as an MDP M = (S, s_I, A, P, R).
- The authors state this as an assumption: We have constructed an abstraction ϕ_M.
- The authors state this as an assumption: The learner accepts elements from S × Q as state input (for the state space of the shield Q).
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, there is no guarantee that unsafe actions are not part of the final policy.Therefore, the shield has to remain active even after the learning phase.”
§ Safe Reinforcement Learning via Shielding > Shielded Reinforcement Learning
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
The authors state this as an assumption.
The specification ϕ is a set of properties {ϕ1,...,ϕl} such that L(ϕ) is the intersection of the languages of its properties.
“We assume ϕ is a set of properties {ϕ 1 , . . ., ϕ l } such that L(ϕ) = i L(ϕ i ).”
The paper relies on this without support.
The environment is often represented as an MDP.
“Given that the environment is often represented as an MDP, such an abstraction has to be conservative w.r.t. the behavior of the real MDP.”
The paper relies on this without support.
Safety or more generally correctness during learning and execution of controllers has attracted significantly less attention than convergence, optimality and data-efficiency of learning algorithms.
“While convergence, optimality and data-efficiency of learning algorithms are relatively well understood, safety or more generally correctness during learning and execution of controllers has attracted significantly less attention.”
The authors state this as an assumption.
The environment can be modeled as an MDP M = (S, s_I, A, P, R).
“Assumption 1 (i) The environment can be modeled as an MDP M = (S, s I , A, P, R).”
The authors state this as an assumption.
We have constructed an abstraction ϕ_M.
“(ii) We have constructed an abstraction ϕ M .”
The authors state this as an assumption.
The learner accepts elements from S × Q as state input (for the state space of the shield Q).
“(iii) The learner accepts elements from S × Q as state input (for the state space of the shield Q).”
The authors state this as an assumption.
Some abstraction ϕ_M = (Q_M, q_0,M, A × L, δ_M, F_M) of M for some MDP observer function f : S → L is given.
“We assume some abstraction ϕ M = (Q M , q 0,M , A × L, δ M , F M ) of M for some MDP observer function f : S → L to be given.”
The authors state this as an assumption.
ϕ_s has Σ = L × A, i.e., it reads the part of the system behavior that the abstraction is concerned with.
“Since ϕ s models a restriction of the traces of the MDP and the learner together that we want to enforce, we assume it to have Σ = L × A, i.e., it reads the part of the system behavior that the abstraction is concerned with.”
The authors state this as an assumption.
A shield S = (Q_S, q_0,S, Σ_I,S, Σ_O,S, δ_S, λ_S) is given for M and for some MDP labeling function f : S → L.
“Assume also that a shield S = (Q S , q 0,S , Σ I,S , Σ O,S , δ S , λ S ) is given for M and for some MDP labeling function f : S → L.”
The authors state this as an assumption.
Whenever an action ranking is chosen by the learner that does not contain a safe action, there is a fixed probability distribution over the safe actions executed instead, constant over time.
“Note that this argument requires that whenever an action ranking is chosen by the learner that does not contain a safe action, there is a fixed probability distribution over the safe actions executed instead.”
The authors state this as an assumption.
The learner has access to the state of the shield and can base its actions on it.
“Note that we made use of the fact that the learner has access to the state of the shield and can base its actions on it in this argument.”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“To provide system-level, physically absolute safety guarantees, SafeCodeRL incorporates a hard shielding mechanism [37].”
From SafeCodeRL: Security-Constrained Multi-Agent Reinforcement Learning for Trustworthy LLM-Generated IoT/CPS Software (2026) · cites this paper as [37]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Safe Reinforcement Learning via Shielding > Synthesis of Shields
“ϕ M = (Q M , q 0,M , A × L, δ M , F M )”
§ Safe Reinforcement Learning via Shielding > Synthesis of Shields
“(δ(q, (l, a)), δ M (q, (l, a))), for all (q, q M ) ∈ G, l ∈ L, a ∈ A”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Safe Reinforcement Learning via Shielding > Synthesis of Shields
“if we merge all states in F × Q M into a single error state”
§ Safe Reinforcement Learning via Shielding > Synthesis of Shields
“F g = (F × Q M ) ∪ (Q × (Q M \ F M )).”
Citations
Cites
Human-level control through deep reinforcement learning
Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A., Riedmiller, M., Fidjeland, A. K., Ostrovski, G., et al. (2015) · Nature · doi:10.1038/nature14236 · Cited by 30,357
Reinforcement Learning: An Introduction
Sutton, R. S., & Barto, A. G. (1998) · IEEE Transactions on Neural Networks · doi:10.1109/tnn.1998.712192 · Cited by 27,225
Reinforcement Learning: An Introduction
Sutton, R. S., & Barto, A. (2005) · IEEE Transactions on Neural Networks · doi:10.1109/tnn.2004.842673 · Cited by 25,268
(2008) · Cited by 4,811
Handbook of theoretical computer science
Leeuwen, J. V. (1990) · Medical Entomology and Zoology · Cited by 4,259
Valiant, L. G. (1984) · ACM eBooks · doi:10.1145/3749740.3749746 · Cited by 4,211
Valiant, L. G. (1984) · Communications of the ACM · doi:10.1145/1968.1972 · Cited by 3,279
Emerson, E. (1991) · Handbook of Theoretical Computer Science, Volume B: Formal Models and Sematics · doi:10.1016/b978-0-444-88074-1.50021-4 · Cited by 3,268
The Arcade Learning Environment: An Evaluation Platform for General Agents
Bellemare, M. G., Naddaf, Y., Veness, J., & Bowling, M. (2013) · Journal of Artificial Intelligence Research · doi:10.1613/jair.3912 · Cited by 2,161
The temporal logic of programs
Pnueli, A. (1977) · 18th Annual Symposium on Foundations of Computer Science (sfcs 1977) · doi:10.1109/sfcs.1977.32 · Cited by 1,523
Cited by
Sorted by citations
Items 201–300 of 700
SafeAPT: Safe Simulation-to-Real Robot Learning Using Diverse Policies Learned in Simulation
Kaushik, R., Arndt, K., & Kyrki, V. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3177294 · Cited by 11
Shielded Reinforcement Learning: A review of reactive methods for safe learning
Odriozola-Olalde, H., Zamalloa, M., & Arana-Arexolaleiba, N. (2023) · IEEE/SICE International Symposium on System Integration (SII) · doi:10.1109/sii55687.2023.10039301 · Cited by 11
SpecGuard: Specification Aware Recovery for Robotic Autonomous Vehicles from Physical Attacks
Dash, P., Chan, E., & Pattabiraman, K. (2024) · ACM SIGSAC Conference on Computer and Communications Security (CCS) · doi:10.1145/3658644.3690210 · Cited by 10
Safe Deep Reinforcement Learning for Resilient Self-Proactive Distribution Grids Against Wildfires
Shalaby, A. A., Abdeltawab, H. H., & Mohamed, Y. A. I. (2025) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2025.3602849 · Cited by 11
Art: Abstraction Refinement-Guided Training for Provably Correct Neural Networks
Lin, X., Zhu, H., Samanta, R., & Jagannathan, S. (2019) · Formal Methods in Computer-Aided Design · doi:10.34727/2020/isbn.978-3-85448-042-6_22 · Cited by 31
Zhong, B., Lavaei, A., Cao, H., et al. (2021) · Nonlinear Analysis Hybrid Systems · doi:10.1016/j.nahs.2021.101110 · Cited by 11
Scukins, E., & Ögren, P. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561853 · Cited by 11
Shielded Deep Reinforcement Learning for Multi-Sensor Spacecraft Imaging
Nazmy, I., Harris, A., Lahijanian, M., & Schaub, H. (2022) · Proceedings of the ... American Control Conference/Proceedings of the American Control Conference · doi:10.23919/acc53348.2022.9867762 · Cited by 11
A UAV Path Planning Method in Three-Dimensional Urban Airspace based on Safe Reinforcement Learning
Li, Y., Zhang, X., Zhu, Y., et al. (2023) · IEEE/AIAA Digital Avionics Systems Conference (DASC) · doi:10.1109/dasc58513.2023.10311219 · Cited by 11
Operation and Imitation Under Safety-Aware Shared Control
Broad, A., Murphey, T. D., & Argall, B. (2020) · Springer proceedings in advanced robotics · doi:10.1007/978-3-030-44051-0_52 · Cited by 11
Automated Safety Verification of Programs Invoking Neural Networks
Christakis, M., Enişer, H. F., Hermanns, H., Hoffmann, J., Kothari, Y., Li, J., Navas, J. A., & Wüstholz, V. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-81685-8_9 · Cited by 12
Systematic Generalisation through Task Temporal Logic and Deep Reinforcement LearningPreprint
León, B. G., Shanahan, M., & Belardinelli, F. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.08767 · Cited by 34
Bo, S., Agyeman, B. T., Yin, X., & Liu, J. (2023) · Computers & Chemical Engineering · doi:10.1016/j.compchemeng.2023.108413 · Cited by 10
Shielded Deep Reinforcement Learning for Complex Spacecraft Tasking
Reed, R. A., Schaub, H., & Lahijanian, M. (2024) · Proceedings of the ... American Control Conference/Proceedings of the American Control Conference · doi:10.23919/acc60939.2024.10644855 · Cited by 10
Niu, Z., Wu, J., & He, H. (2024) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2024.3511032 · Cited by 10
Safe RAN control: A Symbolic Reinforcement Learning Approach
Nikou, A., Mujumdar, A. P., Sundararajan, V., et al. (2022) · International Conference on Control and Automation · doi:10.1109/icca54724.2022.9831850 · Cited by 10
Interpretable, Verifiable, and Robust Reinforcement Learning via Program Synthesis
Bastani, O., Inala, J. P., & Solar-Lezama, A. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-04083-2_11 · Cited by 11
An Abstraction-based Method to Check Multi-Agent Deep Reinforcement-Learning Behaviors
Mqirmi, P. E., Belardinelli, F., & León, B. G. (2021) · Adaptive Agents and Multi-Agents Systems · doi:10.65109/sral5637 · Cited by 12
Formal Specification for Learning-Enabled Autonomous Systems
Bensalem, S., Cheng, C., Huang, X., et al. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-21222-2_8 · Cited by 10
Correct-by-Construction Runtime Enforcement in AI – A Survey
Könighofer, B., Bloem, R., Ehlers, R., & Pek, C. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-22337-2_31 · Cited by 18
Self-citationLearning Transferable Domain Priors for Safe Exploration in Reinforcement Learning
Karimpanal, T. G., Rana, S., Gupta, S., Tran, T., & Venkatesh, S. (2020) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn48605.2020.9207344 · Cited by 11
Faster and Safer Training by Embedding High-Level Knowledge into Deep Reinforcement LearningPreprint
Zhang, H., Gao, Z., Zhou, Y., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.09986 · Cited by 10
Extended Markov Games to Learn Multiple Tasks in Multi-Agent Reinforcement LearningPreprint
León, B. G., & Belardinelli, F. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.06000 · Cited by 10
Scalable Robust Safety Filter With Unknown Disturbance Set
Gruber, F., & Althoff, M. (2023) · IEEE Transactions on Automatic Control · doi:10.1109/tac.2023.3292329 · Cited by 9
Learning Adaptive Safety for Multi-Agent Systems
Berducci, L., Yang, S., Mangharam, R., & Grosu, R. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10611037 · Cited by 11
Learning Collision-free and Torque-limited Robot Trajectories based on Alternative Safe Behaviors
Kiemel, J. C., & Kröger, T. (2022) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids53995.2022.10000077 · Cited by 8
Asarkaya, A. S., Aksaray, D., & Yazıcıoğlu, Y. (2021) · International Conference on Unmanned Aircraft Systems (ICUAS) · doi:10.1109/icuas51884.2021.9476694 · Cited by 9
Dynamic Shielding for Reinforcement Learning in Black-Box Environments
Waga, M., Castellano, E., Pruekprasert, S., et al. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19992-9_2 · Cited by 9
Xiao, W., He, T., Dolan, J. M., & Shi, G. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10611340 · Cited by 13
Liu, D., Brännström, M., Backhouse, A. G., et al. (2019) · IEEE Intelligent Transportation Systems Conference (ITSC) · doi:10.1109/itsc.2019.8917221 · Cited by 9
DSMC Evaluation Stages: Fostering Robust and Safe Behavior in Deep Reinforcement Learning
Gros, T. P., Höller, D., Hoffmann, J., Klauck, M., Meerkamp, H., & Wolf, V. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-85172-9_11 · Cited by 9
Programmatic Reward Design by Example
Zhou, W., & Li, W. (2022) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v36i8.20910 · Cited by 10
A Verification Framework for Certifying Learning-Based Safety-Critical Aviation Systems
Baheri, A., Ren, H., Johnson, B., Razzaghi, P., & Wei, P. (2022) · AIAA AVIATION Forum · doi:10.2514/6.2022-3965 · Cited by 9
Dong, Y., Zhao, X., & Huang, X. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981794 · Cited by 9
Contingency-constrained economic dispatch with safe reinforcement learning
Eichelbeck, M., Markgraf, H., & Althoff, M. (2022) · International Conference on Machine Learning and Applications · doi:10.1109/icmla55696.2022.00103 · Cited by 10
Online Learning of non-Markovian Reward Models
Rens, G., Raskin, J.-F., Reynouard, R., et al. (2021) · International Conference on Agents and Artificial Intelligence (ICAART) · doi:10.5220/0010212000740086 · Cited by 9
Learning Non-Markovian Reward Models in MDPsPreprint
Rens, G., & Raskin, J.-F. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2001.09293 · Cited by 9
Safe Planning via Model Predictive ShieldingPreprint
Bastani, O. (2019) · arXiv (Cornell University) · Cited by 9
PAC Confidence Predictions for Deep Neural Network ClassifiersPreprint
Park, S., Li, S., Lee, I., & Bastani, O. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.00716 · Cited by 9
Formal Specification and Testing for Reinforcement Learning
Varshosaz, M., Ghaffari, M., Johnsen, E. B., & Wąsowski, A. (2023) · Proceedings of the ACM on Programming Languages · doi:10.1145/3607835 · Cited by 9
Zheng, Y., Yan, R., Jia, B., et al. (2024) · Physica A Statistical Mechanics and its Applications · doi:10.1016/j.physa.2024.130137 · Cited by 8
Game-Theoretic Constrained Policy Optimization for Safe Reinforcement Learning
Zhang, C., Zhang, X., Lan, Y., Gao, H., & Xu, X. (2025) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2025.3586603 · Cited by 9
Fang, Z., He, J. Q., Qin, P., Chen, H., Changjian, Z., & Sun, S. (2025) · Accident Analysis & Prevention · doi:10.1016/j.aap.2025.108265 · Cited by 9
Jocas, M., Kurrek, P., Zoghlami, F., Gianni, M., & Salehi, V. (2019) · Proceedings of the ... International Conference on Engineering Design · doi:10.1017/dsi.2019.210 · Cited by 8
Marten: A Built-in Security DRL-Based Congestion Control Framework by Polishing the Expert
Pan, Z., Zhou, J., Qiu, X., et al. (2023) · doi:10.1109/infocom53939.2023.10228952 · Cited by 8
Chriat, A. E., & Sun, C. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3322088 · Cited by 8
Shielded Learning for Resilience and Performance Based on Statistical Model Checking in Simulink
Adelt, J., Bruch, S., Herber, P., et al. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-46002-9_6 · Cited by 8
Safety Filter Design for Neural Network Systems via Convex Optimization
Chen, S., Chee, K. Y., Matni, N., et al. (2023) · Proceedings of the IEEE Conference on Decision & Control, including the Symposium on Adaptive Processes · doi:10.1109/cdc49753.2023.10383577 · Cited by 8
Enhancing Safety in Learning from Demonstration Algorithms via Control Barrier Function Shielding
Yang, Y. F., Chen, L., Zaidi, Z., et al. (2024) · ACM/IEEE International Conference on Human-Robot Interaction (HRI) · doi:10.1145/3610977.3635002 · Cited by 8
Validation of Reinforcement Learning Agents and Safety Shields with ProB
Vu, F., Dunkelau, J., & Leuschel, M. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-60698-4_16 · Cited by 8
Reachability-based Human-in-the-Loop Control with Uncertain Specifications
Gao, Y., Jiang, F. J., Ren, X., et al. (2020) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2020.12.2572 · Cited by 8
A Joint Imitation-Reinforcement Learning Framework for Reduced Baseline Regret
Dey, S., Pendurkar, S., Sharon, G., & Hanna, J. P. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636294 · Cited by 8
Monitor-Based Runtime Assurance for Temporal Logic Specifications
Abate, M., Féron, É., & Coogan, S. (2019) · Proceedings of the IEEE Conference on Decision & Control, including the Symposium on Adaptive Processes · doi:10.1109/cdc40024.2019.9029716 · Cited by 8
CertRL: formalizing convergence proofs for value and policy iteration in Coq
Vajjha, K., Shinnar, A., Trager, B., Pestun, V., & Fulton, N. (2021) · ACM SIGPLAN International Conference on Certified Programs and Proofs (CPP) · doi:10.1145/3437992.3439927 · Cited by 14
Machine Reasoning ExplainabilityPreprint
Čyras, K., Badrinath, R., Mohalik, S. K., Mujumdar, A. P., Nikou, A., Previti, A., Sundararajan, V., & Feljan, A. V. (2020) · arXiv (Cornell University) · doi:10.4230/lipics.cp.2026.62 · Cited by 19
Formal Methods for Autonomous Systems
Wongpiromsarn, T., Ghasemi, M., Cubuktepe, M., Bakirtzis, G., Carr, S., Karabag, M. O., Neary, C., Gohari, P., & Topcu, U. (2023) · Foundations and Trends® in Systems and Control · doi:10.1561/2600000029 · Cited by 17
Self-citationProvable Traffic Rule Compliance in Safe Reinforcement Learning on the Open Sea
Krasowski, H., & Althoff, M. (2024) · IEEE Transactions on Intelligent Vehicles · doi:10.1109/tiv.2024.3400597 · Cited by 7
Odriozola-Olalde, H., Zamalloa, M., Arana-Arexolaleiba, N., & Perez-Cerrolaza, J. (2025) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2025.110055 · Cited by 7
Garg, P., Silvaş, E., & Willems, F. (2025) · Control Engineering Practice · doi:10.1016/j.conengprac.2025.106458 · Cited by 8
Sharifi, I., Yildirim, M., & Fallah, S. (2025) · Transportation Research Record Journal of the Transportation Research Board · doi:10.1177/03611981251357006 · Cited by 24
Guo, S., Zhong, Y., Feng, Z., Kang, S., & Chen, J. (2025) · Journal of Information and Intelligence · doi:10.1016/j.jiixd.2025.12.005 · Cited by 9
Deep reinforcement learning for complex evaluation of one-loop diagrams in quantum field theory
Windisch, A., Gallien, T., & Schwarzlmüller, C. (2020) · Physical review. E · doi:10.1103/physreve.101.033305 · Cited by 7
Safe Robot Navigation Using Constrained Hierarchical Reinforcement Learning
Roza, F. S., Rasheed, H., Roscher, K., Ning, X., & Günnemann, S. (2022) · International Conference on Machine Learning and Applications · doi:10.1109/icmla55696.2022.00123 · Cited by 7
Toward Optimal Live Video Streaming QoE: A Deep Feature-Fusion Approach
Zhang, H., Zhou, A., Wang, G., Li, C., & Ma, H. (2024) · IEEE/ACM Transactions on Networking · doi:10.1109/tnet.2024.3351832 · Cited by 8
Probabilistic reach-avoid for Bayesian neural networks
Wicker, M., Laurenti, L., Patanè, A., et al. (2024) · Artificial Intelligence · doi:10.1016/j.artint.2024.104132 · Cited by 7
Verifying the Generalization of Deep Learning to Out-of-Distribution Domains
Amir, G., Maayan, O., Zelazny, T., Katz, G., & Schapira, M. (2024) · Journal of Automated Reasoning · doi:10.1007/s10817-024-09704-7 · Cited by 8
Formal Verification Techniques for Vision-Based Autonomous Systems – A Survey
Mitra, S., Păsăreanu, C. S., Prabhakar, P., Seshia, S. A., Mangal, R., Li, Y., Watson, C., Gopinath, D., & Yu, H. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-75778-5_5 · Cited by 7
Hartl, B., Levin, M. G., & Zöttl, A. (2025) · Communications Physics · doi:10.1038/s42005-025-02101-5 · Cited by 12
"I’m Sorry Dave, I’m Afraid I Can’t Do That" Deep Q-Learning from Forbidden Actions
Seurin, M., Preux, P., & Pietquin, O. (2020) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn48605.2020.9207496 · Cited by 17
Safe Reinforcement Learning via Probabilistic Timed Computation Tree Logic
Li, Q., & Liu, J. (2020) · doi:10.1109/ijcnn48605.2020.9207384 · Cited by 7
Towards Safe Neural Network Supported Model Predictive Control
Zieger, T., Savchenko, A., Oehlschlägel, T., & Findeisen, R. (2020) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2020.12.1201 · Cited by 8
Reinforcement Learning with Temporal Logic Constraints
Lennartson, B., & Jia, Q. (2020) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2021.04.044 · Cited by 7
Zhang, H., Wang, J., Wu, Z., Wang, Y., & Wang, D. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636519 · Cited by 7
Schmidt, L. M., Rietsch, S., Plinge, A., Eskofier, B. M., & Mutschler, C. (2022) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc55140.2022.9922345 · Cited by 7
Safe reinforcement learning for dynamical systems using barrier certificates
Zhao, Q., Zhang, Y., & Li, X. (2022) · Connection Science · doi:10.1080/09540091.2022.2151567 · Cited by 7
Interactive Imitation Learning in Robotics: A Survey
Celemin, C. E., Pérez‐Dattari, R., Chisari, E., Franzese, G., Rosa, L. D. S., Prakash, R., Ajanović, Z., Ferraz, M., Valada, A., & Kober, J. (2022) · doi:10.1561/9781638281276 · Cited by 7
Reusable Contracts for Safe Integration of Reinforcement Learning in Hybrid Systems
Adelt, J., Brettschneider, D., & Herber, P. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19992-9_4 · Cited by 7
Self-correcting Neural Networks for Safe Classification
Leino, K. R. M., Fromherz, A., Mangal, R., Fredrikson, M., Parno, B. J., & Păsăreanu, C. S. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-21222-2_7 · Cited by 10
Security-Aware Reinforcement Learning under Linear Temporal Logic Specifications
Cui, B., Zhu, K., Li, S., et al. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10160753 · Cited by 7
Hamilton, N. P., Dunlap, K. M., Johnson, T. T., & Hobbs, K. L. (2023) · IEEE International Conference on Space Mission Challenges for Information Technology · doi:10.1109/smc-it56444.2023.00014 · Cited by 9
Towards Cooperative Flight Control Using Visual-Attention
Yin, L., Chahine, M., Wang, T.-H. J., Seyde, T., Liu, C., Lechner, M., Hasani, R., & Rus, D. L. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10342229 · Cited by 8
Certified Vision-Based State Estimation for Autonomous Landing Systems Using Reachability Analysis
Cruz, U. S., & Shoukry, Y. (2023) · Proceedings of the IEEE Conference on Decision & Control, including the Symposium on Adaptive Processes · doi:10.1109/cdc49753.2023.10384107 · Cited by 7
Bypassing the Simulation-to-Reality Gap: Online Reinforcement Learning Using a Supervisor
Evans, B., Betz, J., Zheng, H., Engelbrecht, H. A., Mangharam, R., & Jordaan, H. W. (2023) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar58858.2023.10406465 · Cited by 7
Runtime-Safety-Guided Policy Repair
Zhou, W., Gao, R., Kim, B., Kang, E., & Li, W. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-60508-7_7 · Cited by 7
Learning Optimal Strategies for Temporal Tasks in Stochastic Games
Bozkurt, A. K., Wang, Y., Zavlanos, M. M., et al. (2024) · IEEE Transactions on Automatic Control · doi:10.1109/tac.2024.3390848 · Cited by 7
Safe Data-Driven Contact-Rich Manipulation
Mitsioni, I., Tajvar, P., Kragić, D., et al. (2021) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids47582.2021.9555680 · Cited by 7
Assured Deep Multi-Agent Reinforcement Learning for Safe Robotic Systems
Riley, J., Călinescu, R., Paterson, C., Kudenko⋆, D., & Banks, A. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-10161-8_8 · Cited by 7
Towards Knowledge Infusion for Robust and Transferable Machine Learning in IoT
Fürst, J., Argerich, M. F., Cheng, B., & Kovács, E. (2020) · RonPub -- Research Online Publishing (RonPub UG) · Cited by 7
I'm sorry Dave, I'm afraid I can't do that, Deep Q-learning from forbidden actionPreprint
Seurin, M., Preux, P., & Pietquin, O. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.02078 · Cited by 7
Multi-objective reinforcement learning for designing ethical multi-agent environments
Rodríguez-Soto, M., López-Sánchez, M., & Rodríguez-Aguilar, J. A. (2023) · Neural Computing and Applications · doi:10.1007/s00521-023-08898-y · Cited by 6
Zhu, K., Xue, T., & Zhang, T. (2025) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2025.3578326 · Cited by 10
Constraint-Guided Reinforcement Learning: Augmenting the Agent-Environment-Interaction
Spieker, H. (2021) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn52387.2021.9533996 · Cited by 6
Deontic Paradoxes in ASP with Weak Constraints
Hatschka, C., Ciabattoni, A., & Eiter, T. (2023) · Electronic Proceedings in Theoretical Computer Science · doi:10.4204/eptcs.385.39 · Cited by 6
Reinforcement Learning by Guided Safe Exploration
Yang, Q., Simão, T. D., Jansen, N., Tindemans, S. H., & Spaan, M. T. J. (2023) · Frontiers in artificial intelligence and applications · doi:10.3233/faia230598 · Cited by 6
Haider, T., Roscher, K., Herd, B., et al. (2024) · ACM/SIGAPP Symposium on Applied Computing (SAC) · doi:10.1145/3605098.3635931 · Cited by 6
Control Policy Correction Framework for Reinforcement Learning-based Energy Arbitrage Strategies
Madahi, S. S. K., Gokhale, G., Verwee, M.-S., et al. (2024) · doi:10.1145/3632775.3661948 · Cited by 6
Woodlief, T., Toledo, F., Elbaum, S. G., et al. (2024) · Science of Computer Programming · doi:10.1016/j.scico.2024.103252 · Cited by 6
A Survey of Research on Stability Guarantee of Reinforcement Learning Automatic Control Problem
Gan, Q., Wu, F., & Zhao, J. (2021) · International Conference on Control, Robotics and Cybernetics (CRC) · doi:10.1109/crc52766.2021.9620159 · Cited by 6
Safe Distributional Reinforcement Learning
Zhang, J., & Weng, P. (2022) · Lecture notes in computer science · doi:10.1007/978-3-030-94662-3_8 · Cited by 6
Towards Safe AI: Sandboxing DNNs-Based Controllers in Stochastic Games
Zhong, B.-Z., Cao, H., Zamani, M., & Caccamo, M. (2023) · AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i12.26789 · Cited by 11
Metadata
- Authors
- Alshiekh, M., Bloem, R., Ehlers, R., Könighofer, B., Niekum, S., & Topcu, U.
- Affiliations
- The University of Texas at Austin (US)
- Graz University of Technology (AT)
- University of Bremen (DE)
- Venue
- Proceedings of the AAAI Conference on Artificial Intelligence
- Published
- 2018-04-29
- Volume
- 32
- Issue
- 1
- ISSN
- 2159-5399 (ISSN-L)
- 2374-3468
- OpenAlex
- W2751422670
- Cited by
- 37
- References
- 19
- Citations by year
- 2026: 1 · 2025: 1 · 2024: 3 · 2023: 7 · 2022: 3 · 2021: 7 · 2020: 7 · 2019: 4 · 2018: 4
- FWCI
- 2.24
- Type
- conference paper
- Language
- en
- Open access
- diamond
- doi.org
- Locations
- Proceedings of the AAAI Conference on Artificial Intelligence (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- National Science Foundation (1652113, 1617639, 1646522)
- Austrian Science Fund (S11406-N23, W1255)
- Universität Bremen
- Defense Advanced Research Projects Agency (W911NF-16-1-0001, FA8650-15C)
- Army Research Office (W911NF-15-1-0592, W911NF-15-1-, W911NF-16-1-, W911NF, W911NF-15)
- Air Force Research Laboratory (FA8650-15-C-2546, FA8650-15-, 8650-16-C-2610)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Physical Sciences › Computer Science › Computational Theory and Mathematics › Formal Methods in Verification
- Physical Sciences › Computer Science › Artificial Intelligence › Logic, Reasoning, and Knowledge
- Keywords
- safe reinforcement learning, temporal logic specifications, reactive systems, convergence guarantees, policy learning
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-24, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-24 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
