VLA 算法研究实习生（装载机方向）
Core
Researching Vision-Language-Action (VLA) models for autonomous wheel loader operations, focusing on multimodal data processing, model training, and validation.
Role type
Research Intern (Embodied AI / VLA)
Builds
Multimodal datasets and VLA model prototypes for construction machinery
Domain
Construction equipment / Autonomous robotics
Deliverable
production ML models
Required skills
Python, PyTorch, Transformer architectures, Multimodal learning, VLA model fine-tuning, Reinforcement learning, ROS/ROS 2, Simulation environments
Preferred skills
GR00T/Pi model experience, Model deployment, Distributed training, Paper reproduction
Technologies
PyTorch, ROS, ROS 2, GR00T, Pi series models
Responsibilities
Develop and optimize VLA models for wheel loader tasks; Clean and synchronize multimodal data (vision, language, state, action); Reproduce and adapt VLA models for specific operational scenarios; Evaluate and deploy models in simulation and low-cost hardware platforms; Conduct comparative experiments and failure analysis.
Seniority
Intern
