iGEN
Visit IGEN World Explore IGEN Expo
EXPLORE UPGRADE PLANS
BREAKING
Home ›› Technology ›› Ai ›› Robotics ›› New Training-Free Method Enables Robots to Follow Personalized Commands Like 'Bring My Cup'
iGEN
Technology Artificial Intelligence

New Training-Free Method Enables Robots to Follow Personalized Commands Like 'Bring My Cup'

Researchers propose Visual Attentive Prompting (VAP), a training-free perceptual adapter that enables vision-language-action models to follow personalized commands by using reference images as visual prompts. VAP outperforms generic policies and token-learning baselines on simulation and real-world benchmarks.

iG
iGEN Editorial
July 8, 2026
New Training-Free Method Enables Robots to Follow Personalized Commands Like 'Bring My Cup'

Vision-Language-Action (VLA) models have shown strong generalization to generic instructions, but they struggle with personalized commands such as "bring my cup," where the robot must act on a specific instance among visually similar objects. Researchers from the field of robotics present a new approach called Visual Attentive Prompting (VAP) that addresses this challenge without any additional training.

The Personalization Problem in Robotics

According to the research, VLA models fail to distinguish between a user's personal object and other identical-looking items when given instance-specific instructions. This limitation prevents robots from being deployed in environments where they must handle personalized tasks, such as retrieving a particular cup from a shelf of identical cups.

Visual Attentive Prompting (VAP) Approach

VAP is a simple yet effective training-free perceptual adapter that equips frozen VLA models with top-down selective attention. The method works in three steps:

  • It treats reference images of the user's object as a non-parametric visual memory.
  • It grounds the personal object in the current scene through open-vocabulary detection and embedding-based matching.
  • It injects this grounding as a visual prompt by highlighting the object and rewriting the instruction.

Because VAP requires no retraining, it can be applied directly to existing VLA models, making it practical for real-world deployment where retraining is costly or impossible.

Benchmarks and Experimental Results

To evaluate personalized manipulation, the team constructed two simulation benchmarks—Personalized-SIMPLER and Personalized-VLABench—as well as a real-world tabletop benchmark. These benchmarks test multiple robots across various tasks involving personal objects.

Experiments show that VAP consistently outperforms generic policies and token-learning baselines in both success rate and correct-object manipulation. The method bridges the gap between semantic understanding and instance-level control, delivering reliable performance without any fine-tuning.

Implications for Enterprise Automation

For enterprise technology leaders, this research has direct relevance to warehouse and logistics automation, where robots must pick specific items from bins containing similar-looking products. The training-free nature of VAP means it can be integrated into existing robotics stacks without disrupting operations. While the benchmarks are tabletop-scale, the approach is designed to scale to more complex environments. As VLA models become standard in industrial robots, methods like VAP will be essential for enabling personalized, instance-aware manipulation without retraining costs.

Sources:

Keep Reading

Recommended Stories

RoboSSM Introduces State-Space Models for Scalable In-Context Imitation Learning in Robotics Technology

RoboSSM Introduces State-Space Models for Scalable In-Context Imitation Learning in Robotics

RoboSSM is a new method for in-context imitation learning (ICIL) that replaces Transformer-based architectures with state-space models (SSMs). The approach uses Longhorn, a state-of-the-art SSM, enabling linear-time inference and strong extrapolation to longer prompts. Experiments on the LIBERO benchmark show improved generalization to unseen and long-horizon tasks compared to Transformer-based ICIL methods.

June 20, 2026
New Benchmark and Method Address Occlusion in Vision-Language-Action Models for Robotics Technology

New Benchmark and Method Address Occlusion in Vision-Language-Action Models for Robotics

Researchers introduced LIBERO-Occ, an occlusion-oriented benchmark for Vision-Language-Action (VLA) models, and proposed Viewpoint Imagination (VIM), a method that generates a complementary view from an occluded primary observation to condition action prediction. Experiments show that state-of-the-art VLAs suffer substantial performance degradation under occlusion, and VIM improves robustness across task suites, occlusion types, and severity levels without requiring additional cameras at deployment.

June 16, 2026
MapDream: Task-Driven Map Learning Achieves State-of-the-Art Vision-Language Navigation Technology

MapDream: Task-Driven Map Learning Achieves State-of-the-Art Vision-Language Navigation

Researchers propose MapDream, a framework that learns bird's-eye-view maps directly from navigation objectives rather than hand-crafted reconstruction. The approach achieves state-of-the-art monocular performance on the R2R-CE and RxR-CE benchmarks.

June 16, 2026
LaWAM: Latent World Action Model Enables Efficient, Dynamics-Aware Robot Control with Low Latency Technology

LaWAM: Latent World Action Model Enables Efficient, Dynamics-Aware Robot Control with Low Latency

LaWAM (Latent World Action Model) is a new robotics AI that uses compact latent visual subgoals instead of full video generation to achieve fast, dynamics-aware robot control. It achieves state-of-the-art success rates on LIBERO (98.6%) and RoboTwin (91.22%) with 187ms per action-chunk and up to 24x lower latency than pixel-space World Action Models.

June 16, 2026