This role focuses on algorithm R&D in embodied intelligence. With a solid mathematical foundation, you will conduct research and implementation in one or more of the following areas:
1. Design and optimize Vision-Language-Action (VLA) multimodal large models via linear algebra, probability theory, etc., to map language instructions to physical actions.
2. Build deep learning models for object detection and scene segmentation, and enhance algorithm robustness in dynamic scenarios through mathematical optimization.
3. Develop reinforcement learning frameworks and reward functions for robot navigation and grasping, and narrow the sim-to-real performance gap.
4. Integrate reinforcement learning with traditional control methods based on robot kinematics and dynamics modeling to enable stable robot movement in complex terrains.
Requirements:
1 Master's or PhD graduates from the Universitat Politècnica de Catalunya (UPC)
2 This role requires a strong foundation in mathematical theory and solid programming skills. The candidate will conduct research and development in one or more of the following areas:
* Embodied large-scale models
* Perception algorithms
* Reinforcement learning algorithms
* Motion control algorithms