Mini-BEHAVIOR-Gran benchmark reveals a U-shaped effect of instruction granularity on embodied agent performance, with planning-width correlating best and coarse instructions linked to vision-dominant shallow policies.
DeepThinkVLA: Enhancing Reasoning Capability of Vision-Language-Action Models
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
abstract
Does Chain-of-Thought (CoT) reasoning genuinely improve Vision-Language-Action (VLA) models, or does it merely add overhead? Existing CoT-VLA systems report limited and inconsistent gains, yet no prior work has rigorously diagnosed when and why CoT helps robots act. Through systematic experiments, we identify two necessary conditions that must be jointly satisfied for CoT to be effective in VLA: (1) Decoding Alignment -- CoT and actions must be generated with modality-appropriate mechanisms; forcing both through a single autoregressive decoder is not merely suboptimal but actively harmful, degrading performance by 4.2 percentage points; (2) Causal Alignment -- CoT must be causally linked to task success via outcome-based optimization; without it, supervised CoT is indistinguishable from no reasoning at all under distribution shift, exhibiting a 32.0\,pp performance drop nearly identical to the 31.6\,pp drop of a reasoning-free baseline. Guided by these findings, we build DeepThinkVLA: a hybrid-attention decoder satisfies Condition~1 by pairing causal attention for language with bidirectional attention for parallel action decoding, while a two-stage SFT-then-RL pipeline satisfies Condition~2 by aligning the full reasoning--action chain with sparse task-success rewards. DeepThinkVLA achieves 97.0\% success on LIBERO, 79.0\% robustness on LIBERO-Plus (vs.\ 61.6\% for $\pi_0$-FAST), and 59.3\% success on RoboTwin~2.0, exceeding the strongest baseline by 21.7 points. Furthermore, we validate the practical effectiveness of our approach through real-world robot experiments. Code available at https://github.com/OpenBMB/DeepThinkVLA
years
2026 4verdicts
UNVERDICTED 4representative citing papers
PE-MAMoE combines sparsely gated mixture-of-experts actors with a non-parametric phase controller in MAPPO to maintain plasticity under dynamic user mobility and traffic, yielding 26.3% higher normalized IQM return in simulations.
GazeVLA pretrains on large human egocentric datasets to capture gaze-based intention, then finetunes on limited robot data with chain-of-thought reasoning to achieve better robotic manipulation performance than baselines.
DA-PTQ quantizes VLAs by compensating cross-space distortions and allocating mixed precision to minimize motion errors and kinematic drift in trajectories.
citing papers explorer
-
Mini-BEHAVIOR-Gran: Revealing U-Shaped Effects of Instruction Granularity on Language-Guided Embodied Agents
Mini-BEHAVIOR-Gran benchmark reveals a U-shaped effect of instruction granularity on embodied agent performance, with planning-width correlating best and coarse instructions linked to vision-dominant shallow policies.
-
Plasticity-Enhanced Multi-Agent Mixture of Experts for Dynamic Objective Adaptation in UAVs-Assisted Emergency Communication Networks
PE-MAMoE combines sparsely gated mixture-of-experts actors with a non-parametric phase controller in MAPPO to maintain plasticity under dynamic user mobility and traffic, yielding 26.3% higher normalized IQM return in simulations.
-
GazeVLA: Learning Human Intention for Robotic Manipulation
GazeVLA pretrains on large human egocentric datasets to capture gaze-based intention, then finetunes on limited robot data with chain-of-thought reasoning to achieve better robotic manipulation performance than baselines.
-
DA-PTQ: Drift-Aware Post-Training Quantization for Efficient Vision-Language-Action Models
DA-PTQ quantizes VLAs by compensating cross-space distortions and allocating mixed precision to minimize motion errors and kinematic drift in trajectories.