FineCraft: Enhancing VLA Models for Fine-Grained Manipulation

Shuaike Zhang; Ziheng Quan; Junhao Ran; Muhe Ding; Shaokun Wang; Jianlong Wu

cs.RONov 2, 2026

FineCraft: Enhancing VLA Models for Fine-Grained Manipulation

Authors: Shuaike Zhang, Ziheng Quan, Junhao Ran, Muhe Ding, Shaokun Wang, Jianlong Wu

Organizations: Harbin Institute of Technology, Shenzhen · Xi’an Jiaotong University · Shenzhen Loop Area Institute HITSZ Robotics

Abstract

Fine-grained manipulation couples instruction grounding, local visual precision, and sustained execution. We present FineCraft, a framework that adapts pretrained models for fine-grained manipulation in the MetaFine Challenge. A task-routed mixture of policies selects independently adapted DINOv2-DiT or pretrained $π_{0.5}$ policies through a common observation-action interface. Local visual features help identify the instructed object parts, demonstrations and expert continuations teach the required actions, and new observations guide replanning during execution. FineCraft ranks fourth with an overall score of $0.878$ , exceeding the published reference on all five tasks. Official results and local diagnostics characterize performance across visual perturbations, instruction changes, and multi-stage manipulation. Code is available at https://github.com/eqcy/FineCraft

Figures & tables

Figure 1: FineCraft as a task-routed mixture of policies (MoP). The task ID selects one independent task policy for the rollout. Two backbone families summarize five task instances, including their task-specific branches. The selected policy generates one action stream and replans from new observations.

Task	Goal	$P$	$U$	$B$	Score
T1	Bottle-part grasping	0.787	0.800	0.800	0.796
T2	Mug transport	1.000	1.000	1.000	1.000
T3	Switch toggling	0.925	0.925	0.925	0.925
T4	Cube sorting	0.995	0.983	0.983	0.987
T5	Letter insertion	0.594	0.725	0.725	0.681
Overall (rank 4)		0.860	0.887	0.887	0.878

Table 1: Official five-task results for FineCraft, listed as HITSZ Robotics on the final leaderboard [ 15 ] . Scores use the aggregation in Equation 3 .

Figure 2: Capability gains and local perception diagnostics. (a) Official improvements in $P$ , $U$ , and $B$ over the published $\pi_{0}$ reference, in percentage points (pp) [ 15 ] . (b) Local T1 cap success under camera and lighting perturbations, with $80$ rollouts per condition.

Robotic Manipulation

Robotic manipulation studies how robots perceive, plan, and control physical interactions to grasp, move, assemble, and use objects.

See all 1,736 papers

Improving Fine-Grained Manipulation through Demonstration Expansion and Execution Design
Oct 29, 2026Jingxu Gu, Dongze LianRobotic Manipulation VLA Model Adaptation
Nanjing University
Learning from an Imitable Teacher: Competition Solution for the MetaFine Challenge
Oct 31, 2026Yanze Zhang, Hao Liu, Haochang Hao +3Robotic Manipulation Learning from Demonstration
Beyond Binary Success: A Diagnostic Meta-Evaluation Framework for Fine-Grained Manipulation
May 19, 2026He-Yang Xu, Pengyuan Zhang, Zongyuan Ge +5VLM Evaluation Robotic Manipulation
Southeast University, China · Monash University, Australia · Xiaomi EV, China +2

Oct 29, 2026J/K move · Enter open · S save

Improving Fine-Grained Manipulation through Demonstration Expansion and Execution Design

Jingxu Gu, Dongze Lian

Nanjing University

Robotic Manipulation VLA Model Adaptation Robot Imitation Learning Robustness of VLA Models Closed-Loop Control of VLA Models

Open paper AskPDF arXiv

Fine-grained manipulation requires a robot policy to follow instructions, remain robust to visual variations, and translate predicted actions into precise robot control. We present Demonstration Expansion and Execution Design (DEED), a framework for adapting a pretrained vision-language-action policy to the five MetaFine tasks through structured demonstration expansion and execution design. DEED generates demonstrations for different instructions from the same initial scene and renders each physical trajectory under different camera positions and lighting conditions while preserving its original states and action targets. A hierarchical sampling strategy balances instructions, trajectories, and visual conditions during supervised adaptation. At deployment, DEED converts predicted action chunks into robot commands using inverse kinematics, gripper mappings selected for each task, and closed-loop replanning. DEED achieves an official Overall score of 85.55%. Fixed-policy evaluations show that performance varies with execution horizon, denoising budget, and gripper mapping.

Abstract

Robotic Manipulation

Explore similar work

Improving Fine-Grained Manipulation through Demonstration Expansion and Execution Design

Learning from an Imitable Teacher: Competition Solution for the MetaFine Challenge

Beyond Binary Success: A Diagnostic Meta-Evaluation Framework for Fine-Grained Manipulation