Jindou Jia | NTU Postdoc

Hello, I'm Jindou Jia

I am a postdoc in the MARS Lab at Nanyang Technological University (NTU), working with Prof. Jianfei Yang. I completed my Ph.D. in 2025 from the Shenyuan Honors College, Beihang University, China. I was supervised by Prof. Lei Guo, Prof. Xiang Yu, and Prof. Kexin Guo. My research aims at generalizable robot learning. By integrating world models, generative policies, and uncertainty prediction with physical priors, I work toward robots that reliably transfer from training to unseen environments.

Scholar Github ResearchGate LinkedIn RED

News

Sep 2026: 🎉 ADAPT has been accepted by 2026 CoRL！
Aug 2026: 🎉 PhyFilter has been accepted by Nature npj Robotics！
Aug 2026: 🎉 Our world model survey for robot learning has been accepted by IJRR！
May 2026: ⭐ Released MARS policy, multimodality only when it matters.
Apr 2026: 🎉 A2A policy have been accepted by 2026 RSS！
Feb 2026: 🎉 Two papers have been accepted by 2026 ICRA！
Nov 2025: 🚀 Started my new position as a Postdoc at NTU！
Jul 2025: 🎉 FORESEER has been accepted by IJRR！
Jun 2025: 😊 Passed my PhD doctoral dissertation defense！
Feb 2025: 🎉 Feedback neural network has been selected as an oral paper by 2025 ICLR (top 1.8%)！
Jul 2024: 🎉 Our paper, TRACE, won the Best Student Paper Award at IEEE ICCA 2024!
Oct 2023: 🎉 EVOLVER has been accepted by T-RO！

Robotic Platforms

Hardware platforms I've worked with over the years.

Selected Publications

Highlighted first, co-first^*, or corresponding^† author work. All publications →

★ Featured

arXiv'26

MARS policy: Multimodality only when it matters

Jindou Jia^*, Tuo An^*, Yuxuan Hu^*, Gen Li, Jingliang Li, Bohan Hou, Xiangyu Chen, Jiaqi Bai, Bofan Lyu, Jianfei Yang

arXiv, 2026

We present MARS, a generative policy that adaptively invokes stochasticity only when behavioral branching truly matters and reverts to deterministic learning otherwise, yielding 16.67% higher success and 83.20% lower inference latency in real-world tests.

Project Page PDF arXiv Video

★ Featured

npj Robotics'26

Physics filtering favors the generalization of robot learning

Jindou Jia^*, Shixuan Han^*, Meng Wang^*, Gen Li, Zihan Yang, Sicheng Zhou, Kexin Guo^†, Jianfei Yang^†, Xiang Yu, Wei Wang, Lei Guo

npj Robotics (Nature), 2026

We propose PhyFilter, which enhances both generalization and interpretability of robot learning. PhyFilter operates by correcting learning outcomes according to readily accessible physical differential structure and real-time state feedback.

Link Project Page PDF arXiv Code

★ Featured

RSS'26

Action-to-action flow matching

Jindou Jia^*, Gen Li^*, Xiangyu Chen, Tuo An, Yuxuan Hu, Jingliang Li, Xinying Guo, Jianfei Yang

Robotics: Science and Systems (RSS), 2026

We introduces A2A, an efficient generative paradigm that replaces noise-based initialization with action-to-action transport.

Project Page PDF arXiv Video

★ Featured

ICLR'25

Feedback favors the generalization of neural ODEs

Jindou Jia^*, Zihan Yang^*, Meng Wang, Kexin Guo, Jianfei Yang, Xiang Yu, Lei Guo

International Conference on Learning Representations (ICLR), 2025

Oral Presentation

We present feedback neural networks, showing that a feedback loop can flexibly correct the learned latent dynamics of neural ordinary differential equations (neural ODEs), leading to a prominent generalization improvement.

Link Project Page PDF arXiv Code

★ Featured

T-RO'24

EVOLVER: Online learning and prediction of disturbances for robot control

Jindou Jia, Wenyu Zhang, Kexin Guo, Jianliang Wang, Xiang Yu, Yang Shi, Lei Guo

IEEE Transactions on Robotics (T-RO), 2024

We present a framework, namely EVOLVER, to mimic the bio-behavior for robotics to achieve rapid transient reaction ability and high precision steady-state performance simultaneously.

Link Project Page Code Video

★ Featured

R-AL'22

Accurate high-maneuvering trajectory tracking for quadrotors: A drag utilization method

Jindou Jia, Kexin Guo, Xiang Yu, Weihua Zhao, Lei Guo

IEEE Robotics and Automation Letters (R-AL), 2022

Different from standard approaches that achieve precise tracking by feedforward compensating the estimated drag, this work presents a scheme to appropriately utilize drag.

Link Video

Talks & Media

Interview Robopapers Podcast Guest interview on A2A policy Aug 2026

Invited Talk Seminar on Embodied AI Robotics Efficient robot policy learning · SIMTech & ARTC, Singapore Jun 2026

Invited Talk TechBeat Invited technical talk on feedback neural ODE Mar 2025

Professional Service

Associate Editor IEEE Robotics and Automation Letters (RA-L)

Reviewer IEEE Transactions on Robotics (T-RO), The International Journal of Robotics Research (IJRR), and other robotics journals and conferences

Hello, I'm Jindou Jia

News

Robotic Platforms

Selected Publications

MARS policy: Multimodality only when it matters

Physics filtering favors the generalization of robot learning

Action-to-action flow matching

FORESEER: Recognize and utilize uncertainties by integrating data-based learning and symbolic feedback

Feedback favors the generalization of neural ODEs

EVOLVER: Online learning and prediction of disturbances for robot control

Accurate high-maneuvering trajectory tracking for quadrotors: A drag utilization method

Talks & Media

Professional Service