light_mode

Publications

Papers by Som Sagar on failure diagnosis, red teaming, and evaluation of foundation models, vision-language models, and robot manipulation policies. See also Google Scholar and the CV.

Conference papers

RoboMD probes a trained robot arm policy and returns a ranking of objects by predicted failure probability, across both seen and unseen items.

RoboMD: Uncovering Robot Vulnerabilities through Semantic Potential Fields


Som Sagar, Jiafei Duan, Sreevishakh Vasudevan, Yifan Zhou, Heni Ben Amor, Dieter Fox, Ransalu Senanayake
International Conference on Learning Representations (ICLR), 2026
RoboMD trains a deep reinforcement learning policy to navigate a vision-language embedding space, structured as a potential field of successes and failures, to discover vulnerabilities in robot manipulation policies without real-world trials — up to 23% more unique vulnerabilities than vision-language baselines.
article Details picture_as_pdf arXiv
Humanoid-view scene from PAC Bench with annotated boxes on a pan, the egg inside it, and a sponge — properties and affordances a model must read before acting.

PAC Bench: Do Foundation Models Understand Prerequisites for Executing Manipulation Policies?


Atharva Gundawar*, Som Sagar*, Ransalu Senanayake
Conference on Neural Information Processing Systems (NeurIPS), 2025
PAC Bench is a benchmark of over 30,000 annotations that tests whether vision-language models understand the object Properties, action Affordances, and physical Constraints required for reliable robot manipulation.
article Details picture_as_pdf arXiv
Discover, summarise, restructure: mapping a model's failure landscape, then shifting it with limited human feedback during fine-tuning.

Failures Are Fated, But Can Be Faded: Characterizing and Mitigating Unwanted Behaviors in Large-Scale Vision and Language Models


Som Sagar, Aditya Taparia, Ransalu Senanayake
International Conference on Machine Learning (ICML), 2024 — Spotlight (top 3.5%)
A deep reinforcement learning method that explores and maps the failure landscape of large-scale vision and language models, then restructures that landscape with limited human feedback to mitigate accuracy failures, social biases, and misalignment.
article Details picture_as_pdf arXiv
BaTCAVe answers why a robot's end-effector moved down, scoring human-interpretable concepts with uncertainty bars over the policy network.

Trustworthy Explanations for Robot Behaviors


Som Sagar*, Aditya Taparia*, Harsh Mankodiya, Pranav Bidare, Yifan Zhou, Ransalu Senanayake
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2025
BaTCAVe is a Bayesian concept-based explainability method that produces human-interpretable explanations of robot decisions together with calibrated uncertainty scores, enabling trustworthy post-hoc diagnosis of neural network policies.
article Details picture_as_pdf arXiv
RLPO fine-tunes a diffusion model with LoRA in a reinforcement learning loop, using TCAV preference scores to generate concept images that explain a network.

Explainable Concept Generation through Vision-Language Preference Learning for Understanding Neural Networks' Internal Representations


Aditya Taparia, Som Sagar, Ransalu Senanayake
International Conference on Machine Learning (ICML), 2025
RLPO frames concept-based explanation as an image generation problem and uses reinforcement learning preference optimization to fine-tune a vision-language generative model, automatically producing concept images that reveal what a neural network has internally learned.
article Details picture_as_pdf arXiv
A prompt asks an LLM to convey sadness without saying it; the model replies “the sky seems very grey today” and a grader recovers the intended emotion.

ExpressivityArena: Can LLMs Express Information Implicitly?


Joshua Tint, Som Sagar, Aditya Taparia, Caleb Liu, Kelly Raines, Bimsara Pathiraja, Ransalu Senanayake
Findings of the Association for Computational Linguistics: EACL, 2026
ExpressivityArena is an information-theoretic framework for evaluating how well large language models communicate tone, emotion, identity, and intent implicitly, across nine tasks — revealing that models handle affective content well but lag behind human baselines on sociolinguistic signals.
article Details picture_as_pdf arXiv
Camera poses in 3D: Vantage picks informative viewpoints and avoids occluded views where the cube is hidden and ambiguous ones where 5 cm and 10 cm look alike.

Strategic Vantage Selection for Learning Viewpoint-Agnostic Manipulation Policies


Sreevishakh Vasudevan, Som Sagar, Ransalu Senanayake
IEEE International Conference on Robotics and Automation (ICRA), 2026
Vantage selects a small, strategic set of camera poses for fine-tuning, formulating camera placement as an information-gain optimization so that pre-trained manipulation policies become robust to viewpoint shifts at deployment — raising task success by 25% for diffusion policies.
article Details picture_as_pdf arXiv

Workshop papers

A diagnostic agent probes a trained manipulation policy and ranks objects by predicted failure probability across seen and unseen items.

From Mystery to Mastery: Failure Diagnosis for Improving Manipulation Policies


Som Sagar, Jiafei Duan, Sreevishakh Vasudevan, Yifan Zhou, Heni Ben Amor, Dieter Fox, Ransalu Senanayake
Robotics: Science and Systems (RSS) Workshop on Out-of-Distribution Generalization in Robotics, 2025
A deep reinforcement learning framework that systematically diagnoses failure modes of robot manipulation policies under unseen environmental variations, and uses the discovered vulnerabilities to fine-tune and improve policy robustness.
article Details picture_as_pdf arXiv
GPT-4o scores the mismatch between a prompt and the diffusion model's generated panda image, producing the reward signal that drives red teaming.

LLM-Assisted Red Teaming of Diffusion Models through “Failures Are Fated, But Can Be Faded”


Som Sagar, Aditya Taparia, Ransalu Senanayake
NeurIPS Workshop on Red Teaming GenAI: What Can We Learn from Adversaries?, 2024
An extension of the Failures Are Fated framework to text-to-image diffusion models with LLM-generated rewards and states, action screening inspired by design of experiments, and a comparison of DQN, PPO, and A2C for red teaming.
article Details picture_as_pdf arXiv

About the author. Som Sagar is a computer science PhD student at Arizona State University, advised by Ransalu Senanayake in the LENS Lab. He works on reinforcement learning for post-training and agentic LLM systems, and on failure diagnosis, red teaming, and safety evaluation of foundation models, vision-language models, and robot manipulation policies.

home Homepage library_books All publications description CV school Google Scholar