Jiawei Xia

I am an incoming master's student at the University of California, Berkeley. My research interests include Robot manipulation and Spatial intelligence.

Previously, I worked as a research intern at Institute for AI Industry Research, Tsinghua University and an Embodied AI Algorithm Intern at JD Explore Academy. And I enjoyed my time there.

In my spare time, I enjoy photography, hiking, formula 1, and basketball.

✉️Email:  xiaarrold@berkeley.edu

LinkedIn  /  X  /  Rednote  /  YouTube   

🔥 News

• [2026.03] 🎉 I joined JD Explore Academy as an Embodied AI Algorithm Intern, Beijing, China.

• [2025.07] 🎉 I was honored to receive the only First Prize in the Embodied AI Entrepreneurship Bootcamp organized by HKCRC, Hong Kong, China.

• [2025.04] 🏆 Our team won Third Prize in the DJI Technology RoboMaster University League Competition 2025, Shandong, China.

• [2024.12] 🎉 I joined the Institute for AI Industry Research, Tsinghua University as an Intern, Beijing, China.

• [2024.11] 🎉 I was honored to receive the Xiaomi Scholarship.

• [2024.06] 🏆 Our team won Second Prize in RoboCup@Home 2024, Fujian, China.

🧑‍🎓 Education

💻 Internship

Responsible for hardware debugging, data acquisition system development, model training and deployment based on the ARX5 ALOHA dual-arm robot platform.

Participated in the research on reinforcement learning post-training in robotic VLA models and the development of 3D/4D scene models for embodied intelligence.

Responsible for the development of the Robust LiDAR-Inertial-Thermal SLAM system and other Visual-Inertial Odometry.

Participated in the research of a multimodal framework for robotic kitchen manipulation, including the integration of tactile and thermal imaging sensors into robot imitation learning.

📚 Publications

* : Equal contribution; †: Corresponding author(s)

While world models evaluate future states well, effective continuous control relies on generating good candidate actions. Existing planners either search arbitrarily or require bloated architectures (e.g., VLMs) for action priors. We introduce PRISM, a simple, task-agnostic framework. Attached to a frozen JEPA encoder, a lightweight MLP predicts a Gaussian action prior. At plan time, a closed-form Product-of-Gaussians update integrates this prior to guide sampling based on confidence.

Thermal imaging handles adverse conditions but suffers from AGC and low contrast, violating brightness constancy. To solve this, we propose PL-LIT, a general LiDAR-Inertial-Thermal SLAM system compatible with both visible and thermal cameras. Built on an ESIKF framework, it integrates online photometric calibration with deep point-line feature extraction and robust line constraints. Furthermore, it constructs a probabilistic thermal voxel map to support real-time anomaly detection.

This paper proposes a consistent and tightly-coupled visual-inertial-ranging odometry system based on the Lie group. Our method incorporates the UWB anchor state into the system state, explicitly accounting for UWB calibration uncertainty and enabling the joint and consistent estimation of both robot and anchor states.

This paper proposes a robust online calibration method for UWB anchors in Visual-Inertial Navigation Systems. Based on MSCKF, our approach uses stochastic optimization for reliable initialization and a Schmidt Kalman Filter for online refinement.

🤖 Projects

Our approach is based on fine-tuning the pre-trained Vision-Language-Action (VLA) model, pi0.5. To tackle the competition tasks, the team explored efficient post-training data utilization strategies, including data frequency optimization, DAgger data collection, sub-task segmented prompting, and advantage data filtering. During inference, action smoothing was applied. Ultimately, the model achieved a success rate of over 85% on the deformable/flexible clothing tasks.

This project tried two ways about reinforcement learning post-training on robotic models. Firstly, we used the DAgger framework and positive/negative labeling to conduct rl post-training on the vision language action model, enabling the robot to perform the clothes-folding task. Secondly, we leverage sparse task success rewards and an off-policy actor-critic framework to fine-tune policy in simulation environment.

Developed an intelligent home service robot using a Kinova Gen2 arm and an Azure Kinect camera on ROS Noetic, with functions including guest reception and guidance, voice interaction, object recognition, and room cleaning. Used Gemini, SAM2, and GraspNet to estimate object pose and grasp objects. Controlled the arm using Actionlib and MoveIt. Applied FAST-LIO2 with Livox MID-360 for indoor mapping and localization.

Based on Kongfubot, used inverse kinematics retargeting and reinforcement learning to enable humanoid robots to imitate highly-dynamic behaviors (e.g., kungfu, dancing). Developed sim2sim in MuJoCo and sim2real with Unitree SDK, successfully deploying the framework on the Unitree G1 robot.

Used PnP to calculate the 3D position of the armor plate, and used Kalman filtering for tracking; The robot used four M3508 motors for the chassis and two M6020 motors for the turret, controlled via CAN communication and PID algorithms for current, speed, and position loops. Used feedforward PID to improve the gimbal drive and achieve gravity compensation.

ICRA 2024 RoboMaster Sim2real Challenge: In Habitat sim environment in docker provided by organizers, utilized ArUco for detecting poses of boxes, extracted the transformation and rotation matrices of box from solvePnP. Deployed an Extended Kalman Filter alongside an omnidirectional motion model for state estimation using sensor data, including IMU, odometry and depthimage to laserscan.

🏆 Honours & Awards

• 2026 Beijing Outstanding Undergraduate Graduate
• 2025 First Prize in the Embodied AI Entrepreneurship Bootcamp organised by Hong Kong Center for Construction Robotics
• 2025 Third Prize in DJI Technology RoboMaster University League Competition 2025, China, Shandong Province
• 2024 Xiaomi Scholarship
• 2024 Second Prize in Robocup@Home Competition (leader)
• 2024 Second Prize in China iCAN Innovation Contest (leader)
• 2023 Beijing University of Chemical Technology Excellent Student Leader
• 2021 First Prize in DJI Technology RoboMaster Youth East China League Competition

🏀 Leadership & Activities

• Leader of the SIE Robotics Club
• Signed Photographer for 500px and Visual China
• Leader of the BUCT EYE Photography Club
• Leader of the Comprehensive Office, BUCT Student Union
• Member of BUCT SIE Basketball Team
• Teaching Assistant of Robotic Workshop at BUCT

Design and source code from Jon Barron's website.