Tag
本文提出 Code-Only-As-Policy (COAP) 范式，将机器人本体视为图灵机，用显式代码替代 VLM/VLA 进行决策与状态跟踪，并借助编码智能体实现闭环递归自我改进；在 RoboDojo 的 42 项双臂任务上达到 70.24% 成功率，且测试时无需模型在环。
PhysCaP is a physics-informed code-generation agent that actively explores objects to infer hidden physical properties for efficient robotic manipulation.
A survey paper organizing robot-learning techniques along an axis of frozen-weight policies (VLA models) versus agents that write their own executable skills as code, providing a taxonomy of self-improvement mechanisms and analyzing the emerging robot-skill economy.
MEMENTO introduces a memory-guided memetic framework for evolving robot control programs (code-as-policy), outperforming existing methods like Eureka and REvolve in long-horizon embodied tasks, and demonstrating sim-to-real transfer.
ASPIRE is a continual learning system that autonomously develops and refines robot control programs through iterative exploration, achieving significant improvements in manipulation and household tasks while enabling sim-to-real transfer.
Introduces Playful Agentic Robot Learning, where embodied coding agents use self-directed play to learn reusable skills, improving downstream task performance without additional training. The proposed RATs system achieves significant gains over baselines in simulation and real-world transfer.
The author gave their OpenClaw AI agent a physical robot arm, successfully configuring it to grab objects and train another AI model, demonstrating that AI can simplify robotics control.