AI · Robotics · Computer Vision
Technical Team Lead — Autonomous Subsea Robotics
I build physical AI for the ocean floor — agentic decision-making, learned behaviours, and semantic perception fused with classical robotics — so machines can inspect critical infrastructure at depth, without a human at the controls.
$ whoami Pep Lluís Negre Technical Team Lead — Robotics & AI $ cat focus.txt Agentic Decision-Making Learned Behaviours (VLA) · Semantic Perception SLAM · Sensor Fusion · Stereo Vision $ ./career --highlights SubSLAM — live 3D mapping World-first jacket inspection 6 patents · 4 publications
About
I'm a robotics & AI engineering leader with 14+ years building production-grade autonomous systems for harsh, real-world environments — underwater, unsupervised, and at depth.
My current focus is physical AI for subsea robotics — agentic decision-making, learned behaviours (VLA models), and semantic perception — setting technical strategy and leading a cross-functional team for Rosenxt's autonomous subsea inspection programme. Previously drove two world-leading robotics technologies: SubSLAM, a live 3D mapping system for subsea infrastructure, and the world-first autonomous offshore jacket inspection system.
Still hands-on with the classical foundations — SLAM, stereo vision, sensor fusion — that underpin that autonomy, with extensive experience in hardware-software integration and cross-functional collaboration with mechanical, electrical, and operations teams.
Experience
Setting technical strategy and leading a cross-functional team for Rosenxt's autonomous subsea inspection programme, with a current focus on physical AI — agentic decision-making, learned behaviours, and semantic perception — alongside architecture, hiring, and delivery.
Delivered SubSLAM's real-time 3D stereo-vision system and the world-first subsea autonomous inspection algorithm, combining vision, mapping, and ML.
Led the world-first autonomous inspection of an offshore jacket structure — architecture, sensor fusion, decision-making, and mission autonomy.
Built Rovco's live 3D vision for ROVs and AUVs, improving camera pose accuracy and dense 3D reconstruction in challenging underwater imagery.
Designed and delivered SubSLAM — Rovco's online 3D mapping sensor for subsea infrastructure.
Co-built the autonomous underwater robot "Turbot" — localization, stereo vision, loop-closing, and 3D reconstruction.
+ earlier roles in software and control engineering (2009 – 2015) — see the full CV ↗
Projects & Patents
AI-driven autonomous inspection of an offshore jacket structure — a world-first for subsea infrastructure.
Real-time stereo-vision 3D mapping sensor for subsea infrastructure inspection.
Stereo camera SLAM system from my ICRA'16 paper — Best Student Paper Award finalist.
Loop-closure detection system based on my Autonomous Robots journal paper.
Contributions to viso2 and fovis, two of ROS's most widely used visual odometers.
Drivers for industrial stereo cameras used across the SRV group's AUV platforms.
Skills
Based in Spain, working remote-only. Always happy to talk subsea autonomy, underwater perception, or scaling robotics teams.