← Back to Benchmarks
simmediumroboticsmetric · varies

Towards Generalizable Robotic Manipulation in Dynamic Environments

Description

Vision-Language-Action (VLA) models excel in static manipulation but struggle in dynamic environments with moving targets. This performance gap primarily stems from a scarcity of dynamic manipulation datasets and the reliance of mainstream VLAs on single-frame observations, restricting their spatiotemporal reasoning capabilities. To address this, we introduce DOMINO, a large-scale dataset and benchmark for generalizable dynamic manipulation, featuring 35 tasks with hierarchical complexities, ove

Source

http://arxiv.org/abs/2603.15620v1