SafeBranch: Branch-Pair Safety Alignment for Embodied Agents
cs.AI, cs.CV, cs.RO
Submitted: 2026-08-20
Updated: 2026-10-13
Terminology
Sources
- Anchored Preference Optimization and Contrastive Revisions: Addressing Underspecification in Alignment
- VLA-Reasoner: Empowering Vision-Language-Action Models with Reasoning via Online Monte Carlo Tree Search
- TCPO: Thought-Centric Preference Optimization for Effective Embodied Decision-making
- RoboMonkey: Scaling Test-Time Sampling and Verification for Vision-Language-Action Models
- FailSafe: Reasoning and Recovery from Failures in Vision-Language-Action Models
- IS-Bench: Evaluating Interactive Safety of VLM-Driven Embodied Agents in Daily Household Tasks
- HomeGuard: VLM-based Embodied Safeguard for Identifying Contextual Risk in Household Task
- C-MCTS: Safe Planning with Monte Carlo Tree Search
- Direct Preference Optimization: Your Language Model is Secretly a Reward Model
- Safety Guardrails for LLM-Enabled Robots
- CHOP: Counterfactual Human Preference Labels Improve Obstacle Avoidance in Visuomotor Navigation Policies
- SafetyALFRED: Evaluating Safety-Conscious Planning of Multimodal Large Language Models
- World Modeling Makes a Better Planner: Dual Preference Optimization for Embodied Task Planning
- Tree of Thoughts: Deliberate Problem Solving with Large Language Models
- SafeAgentBench: A Benchmark for Safe Task Planning of Embodied LLM Agents
- SafeVLA: Towards Safety Alignment of Vision-Language-Action Model via Constrained Learning
- GRAPE: Generalizing Robot Policy via Preference Alignment
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems