WoRV Team is Hiring!

(Open) Research Engineer (Spatial AI)

Role Overview — Research Engineer (Spatial AI)

“Researching and building AI systems that understand space and infer location like humans do.”

The WoRV (World model for Robotics and Vehicle control) team’s Spatial AI Research Engineers develop technologies that enable robots deployed in industrial environments to localize themselves and build maps using camera sensors. Our goal is to research and develop models that can reason about their surroundings and infer their own location like humans.

The Spatial AI Research team is currently focused on three core missions:

1. Human-like Localization
   • Develop a system that allows robots to infer their location from abstract maps or environmental cues, similar to how humans use sketches or schematic maps.
   • We are researching human-like localization based on OrienterNet and building an occupancy-map-based pose estimation system.
2. Human-Like Exploration
   • Research model-intrinsic Mapping + Localization + Navigation systems that can understand the environment and navigate without pre-built maps.
   • Go beyond simple pose estimation to build Spatial AI capabilities that comprehend structure and semantics of the environment and use them for action.
3. End-to-End SLAM
   • Move past the limits of modular, rule-based SLAM by developing end-to-end SLAM with a single integrated model.
   • Integrate Dense SLAM, Foundation-Model-based localization, and Visual Localization & Relocalization to build a more robust and accurate system.

Research Support

• Ultra-High-Performance GPU Cluster: CORE (Compute-Oriented Research Environment) [CORE Introduction]

  • On-premise DGX H100: 12 units (96×H100), 30+ A100s, and 10+ V100s in operation for research.

• High-quality, fully human-driven driving-data pipeline for both simulation and real-world environments

  • Unlimited custom datasets and annotations provided to meet research needs.

• Extensive map-based driving data

  • Over 200 hours available, with unlimited access for the research team.

• Full support for conferences & publications

  • NeurIPS, ICLR, CVPR, ECCV, IROS, ICRA, etc.
  • Full coverage of conference participation expenses upon acceptance to international venues.

  WoRV Tech Blog

  Publications | maum.ai BRAIN Team

  Open-Source Activities | maum.ai BRAIN Team

Key Responsibilities

• Human-like Localization R&D
  • Build human-level localization using state-of-the-art methods such as OrienterNet and map-relative pose regression.
  • Develop robust pose estimation using abstract map information (e.g., sketch maps, occupancy maps).
• Foundation-Model-based Spatial Understanding
  • Research FM-based localization (e.g., FM-Loc, AnyLoc, FoundLoc).
  • Develop semantic mapping and spatial understanding with VLMs (Vision-Language Models).
• Experimentation & Validation in Simulation and the Real World
  • Evaluate and optimize performance across diverse indoor/outdoor environments.
  • Tailor technology to industrial requirements and prepare for commercialization.

Required Qualifications

• A strong drive to proactively define problems and deliver solutions.