My research sits at the intersection of Robotics and Machine Learning, with a focus on scaling robot learning & motion planning to long-horizon tasks, high-dimensional state spaces, large plan sets, and multimodal (often non-convex) solution landscapes. I’m especially interested in vectorized/batched planning and optimization (e.g., tensorized search) and in combining algorithmic structure with modern generative models (diffusion/flow-matching) to build planners and policies that are efficient, robust, and deployable, recently with a strong emphasis on humanoid locomotion and manipulation.

Outside of robotics, I’m also a hobbyist in numerical physics/general relativity: I build simulation code to study dynamical processes in curved spacetime (e.g., Kerr black holes and the Penrose process, spacetime energy condition verifications, etc.), and I enjoy the “physics-style” workflow of modeling, validation, and reproducible computational experiments.

• Model Tensor Planning - ICLR 2026 / TMLR 2025
• Global Tensor Motion Planning - ICRA 2026 / IEEE RA-L 2025
• Motion Planning Diffusion - IROS 2023 / IEEE T-RO 2025 / AAAI 2026
• Accelerating Motion Planning via Optimal Transport - NeurIPS 2023

• Penrose process in Kerr spacetime:
  • arXiv: https://arxiv.org/abs/2601.19616
  • code: https://github.com/anindex/penrose_process
• Observer-robust energy condition verification for warp drive spacetimes:
  • arXiv: https://arxiv.org/abs/2602.18023
  • code: https://github.com/anindex/warpax

Vectorized robot learning & planning • Tensor search / batched optimization • OT & gradient flows • Diffusion / flow matching for motion generation • Humanoid locomotion & loco-manipulation • VLA/VLM for grasping & manipulation • Reproducible simulation + numerical GR

• 🌐 Website: anthaile.com
• 💌 Email: an@robot-learning.de
• 🐦 Twitter/X: @an_thai_le