Emergent Active Perception and Dexterity of Simulated Humanoids from Visual Reinforcement Learning

TL;DR

This paper presents Perceptive Dexterous Control (PDC), a vision-based framework enabling simulated humanoids to perform complex household tasks through active perception and reinforcement learning, without relying on privileged state information.

Contribution

Introduction of PDC, a novel vision-driven control framework allowing simulated humanoids to perform multiple tasks using egocentric vision and reinforcement learning, with emergent active search behaviors.

Findings

PDC enables humanoids to perform object search, grasping, and manipulation tasks.

Reinforcement learning from scratch produces human-like active search behaviors.

The approach demonstrates the importance of perception-action loops in embodied AI.

Abstract

Human behavior is fundamentally shaped by visual perception -- our ability to interact with the world depends on actively gathering relevant information and adapting our movements accordingly. Behaviors like searching for objects, reaching, and hand-eye coordination naturally emerge from the structure of our sensory system. Inspired by these principles, we introduce Perceptive Dexterous Control (PDC), a framework for vision-driven dexterous whole-body control with simulated humanoids. PDC operates solely on egocentric vision for task specification, enabling object search, target placement, and skill selection through visual cues, without relying on privileged state information (e.g., 3D object positions and geometries). This perception-as-interface paradigm enables learning a single policy to perform multiple household tasks, including reaching, grasping, placing, and articulated object manipulation. We also show that training from scratch with reinforcement learning can produce emergent behaviors such as active search. These results demonstrate how vision-driven control and complex tasks induce human-like behaviors and can serve as the key ingredients in closing the perception-action loop for animation, robotics, and embodied AI.