Industrial embodied AI
Building intelligent systems for the physical world.
Researching small deployable AI models, vision-language systems, robotics, and autonomous manufacturing intelligence.
A compact AI layer that observes a manufacturing cell, reasons about the next process move, and coordinates the machines already on the floor.
Read cell state from vision, robot pose, tools, and part context.
Choose the next process step with constraints and uncertainty in mind.
Send compact decisions to existing robots, machines, and operators.
Small, deployable, cell-aware
Existing robots and machines
About
David Seyser works at the intersection of AI, robotics, and industrial autonomy, with a focus on models that can leave the lab and operate inside physical workflows.
His work spans embodied AI, robot-based machining, manufacturing intelligence, agriculture automation, vision-language models, LLMs, planning, and representation learning for systems that need to perceive, decide, and coordinate action under real constraints.
Research focus
Learning systems that connect perception, action, and feedback inside physical environments.
Multimodal models for interpreting state, context, and visual process signals.
Efficient deployable models designed for constrained and specialized industrial settings.
AI coordination layers for cells, machines, robots, operators, and process plans.
Planning and control for robotic manufacturing processes with stiffness and process constraints.
Internal state models that support long-horizon decisions and adaptable autonomy.
Featured vision
The long-term goal is to build AI systems that perceive manufacturing cell states, reason about next process steps, and coordinate existing robots and machines without requiring every factory to be rebuilt from scratch.
Projects
Placeholder project areas for research, prototypes, and applied systems moving toward industrial embodied AI.
A coordination layer for perceiving manufacturing cell state and choosing process-aware next actions.
Representation and regularization work for more stable visual encoders in downstream systems.
Compact language models optimized for specialized reasoning and edge-adjacent deployment.
Planning infrastructure for simulated and real manufacturing cells with coordinated resources.
Applied autonomy experiments for sensing, decision-making, and robotic assistance in agriculture.
Contact
Reach out for research conversations, collaborations, applied prototypes, or founder-track work in embodied AI and robotics.