From Intention to Execution: Probing the Generalization Boundaries of Vision-Language-Action Models
read the original abstract
One promise that Vision-Language-Action (VLA) models hold over traditional imitation learning for robotics is to leverage the broad generalization capabilities of large Vision-Language Models (VLMs) to produce versatile, "generalist" robot policies. However, current evaluations of VLAs remain insufficient. Traditional imitation learning benchmarks are unsuitable due to the lack of language instructions. Emerging benchmarks for VLAs that incorporate language often come with limited evaluation tasks and do not intend to investigate how much VLM pretraining truly contributes to the generalization capabilities of the downstream robotic policy. Meanwhile, much research relies on real-world robot setups designed in isolation by different institutions, which creates a barrier for reproducibility and accessibility. To address this gap, we introduce a unified probing suite of 50 simulation-based tasks across 10 subcategories spanning language instruction, vision, and objects. We systematically evaluate several state-of-the-art VLA architectures on this suite to understand their generalization capability. Our results show that while VLM backbones endow VLAs with robust perceptual understanding and high level planning, which we refer to as good intentions, this does not reliably translate into precise motor execution: when faced with out-of-distribution observations, policies often exhibit coherent intentions, but falter in action execution. Moreover, finetuning on action data can erode the original VLM's generalist reasoning abilities. We release our task suite and evaluation code to serve as a standardized benchmark for future VLAs and to drive research on closing the perception-to-action gap. More information, including the source code, can be found at https://ai4ce.github.io/INT-ACT/
This paper has not been read by Pith yet.
Forward citations
Cited by 8 Pith papers
-
MANGO: Automated Multi-Agent Test Oracle Generation for Vision-Language-Action Models
MANGO uses Generator, Assessor, and Judge agents to create reusable atomic tasks and fine-grained oracles from natural language, evaluated on LIBERO_10 and RoboCasa benchmarks for comparable failure detection with bet...
-
APT: Action Expert Pretraining Improves Instruction Generalization of Vision-Language-Action Policies
APT pretrains the action expert as a vision-action prior on frozen VLM features then adds language through gated fusion to improve OOD instruction generalization in continuous-action VLA policies.
-
EmbodimentSemantic: A Spatial Scene-Graph Dataset and Benchmark for Vision-Language Models on Embodied Manipulation Trajectories
EmbodimentSemantic is a spatial scene-graph dataset and benchmark for evaluating relational grounding in vision-language models on embodied manipulation trajectories.
-
AxisGuide: Grounding Robot Action Coordinate System in RGB Observations for Robust Visuomotor Manipulation
AxisGuide augments RGB images with rendered robot base-frame axis cues to improve generalization of visuomotor manipulation policies under distribution shifts.
-
RoboSemanticBench: Diagnosing Semantic Grounding in Action Prediction for VLA Models
RoboSemanticBench reveals that representative VLA models grasp blocks successfully but select the semantically correct answer at near-random rates, indicating a gap between backbone semantics and action prediction.
-
Seeing Realism from Simulation: Efficient Video Transfer for Vision-Language-Action Data Augmentation
A video transfer pipeline augments simulated VLA data into realistic videos while preserving actions, yielding consistent performance gains on robot benchmarks such as 8% on Robotwin 2.0.
-
Genie Sim 3.0 : A High-Fidelity Comprehensive Simulation Platform for Humanoid Robot
Genie Sim 3.0 introduces an LLM-powered scene generator, the first LLM-based automated evaluation benchmark, and a large open synthetic dataset that demonstrates zero-shot sim-to-real transfer for robotic manipulation...
-
LIBERO-PRO: Towards Robust and Fair Evaluation of Vision-Language-Action Models Beyond Memorization
LIBERO-PRO shows VLA models collapse from over 90% to 0% accuracy under perturbations in objects, states, instructions, and environments, exposing memorization instead of genuine comprehension.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.