Boxuan Ye
[email protected] | +44 7422 720280 | ORCID: 0009-0004-3104-2788 | yeboxuan.com
Education
Imperial College London · June 2023 - June 2027
MEng Design Engineering (four-year integrated Bachelor’s and Master’s) | First Class (expected) | London, UK
Selected coursework: Machine Learning, Robotics, Computing, Data Science, Electronics and Finite Element Analysis.
Publications
B. Ye and D. Boyle. TRUCE: Route-Guided Recovery with Predictive Continuation for Quadrotor Navigation. Submitted to ICRA 2027.
P. Hou, Y. Zhang and B. Ye. Real-Gen: Real-Scene-Grounded Generative Augmentation for 6D Object Pose Estimation. Expert Systems with Applications, 332, 133619 (online 2026; issue 2027).
P. Hou, Y. Zhang, B. Ye and Y. Wu. Bridging the Depth Gap: Adaptive Scene-Instance Alignment for Training-Free Depth Refinement in Robotic Manipulation Scenes. IEEE Robotics and Automation Letters, 11(5), 5454-5461, 2026.
P. Hou, Y. Zhang, W. Zhou, B. Ye and Y. Wu. A lightweight network for category-level open-vocabulary object pose estimation with enhanced cross implicit space transformation. Engineering Applications of Artificial Intelligence, 155, 110890, 2025.
Research Experience
UAV Teleoperation and Robotic Manipulation
Penn State University | Research Assistant · July 2026 - Present
Aerial Robots & Interaction (ARI) Lab | Advisor: Junyi Geng
Configure operator viewpoints and camera views for UAV teleoperation.
Develop robotic-arm control functionality and software simulations for teleoperation experiments.
Physics-Constrained Wildfire Spread and Fireline Prediction
Texas A&M University | Research Assistant · July 2026 - Present
Advisor: Yiwei Lyu
Develop fireline forecasts combining learned spread-rate corrections with differentiable anisotropic Eikonal propagation.
Build FireBench and WRF-SFIRE simulation and evaluation pipelines; compare fireline geometry and arrival-time errors against physical and neural baselines.
Investigate fire-intensity prediction and uncertainty modelling with fuel-consumption and energy-budget constraints.
TRUCE: Route-Guided Recovery with Predictive Continuation for Quadrotor Navigation
Imperial College London | Individual Research Project · April 2026 - Present
Systems and Algorithms Laboratory | Supervisor: Dr David Boyle
Developed TRUCE, coupling persistent route guidance, recovery from measured vehicle state and predictive continuation checks around existing quadrotor planners; checked polynomial references using Bernstein bounds and subdivision, with separate checks of predicted vehicle motion.
Achieved 83% goal reaching with no recorded contacts on 100 shared MetaUrban tasks under simulated quadrotor dynamics, versus 22% for official ROS EGO v1 and 26% for EGO v1 with Gatekeeper.
Integrated TRUCE with EGO v1, EGO v2, MINCO and SANDO; a separate 240-run simulation study increased aggregate goal reaching from 104/120 to 118/120 and reduced recorded contacts from eight to zero, at increased computation and travel time.
Mixed-Reality Teleoperation and Open-Vocabulary 6D Pose Estimation
Central Research Centre, Midea Group | Research Assistant · July 2025 - October 2025
Developed Vision Pro humanoid teleoperation with spatial anchoring, hand/object tracking, bidirectional streaming and recording/playback for manipulation and reinforcement-learning data collection.
Integrated GroundingDINO, SAM-HQ, DINOv3 and FoundationPose for CAD-based and non-CAD object pose estimation.
Led electrical control, software architecture and inter-module communication for an oven-based 3D printer; debugged perception, control and visualization integration.
Multi-View Data Generation and Pose Annotation
Blender Synthetic Data Toolchain | Independent Project · July 2025 - August 2025
Built a Blender Python pipeline for RGB, depth, segmentation and pose annotations with camera trajectories, domain randomization and viewpoint perturbations.
Integrated Pyrender-based simulation, PLY reconstruction, automatic oriented bounding boxes and multi-modal export for retrieval and pose estimation.
Vision-Based Perception and Manipulation for Household Robots
Embodied Intelligence, KUKA Robotics | RA · July 2024 - September 2024
Implemented oven-door and screw-tightening tasks with stereo vision; built RoboCasa, Isaac Gym and MuJoCo reinforcement-learning pipelines with custom rewards, domain randomization and sim-to-real transfer.
Developed ROS monitoring and robot-state displays; contributed to 6D pose-estimation algorithms, experiments and manuscript writing.
Six-Degree-of-Freedom Manipulation
Robotic Arm Grasping | Independent Project · July 2024 - September 2024
Combined Python/UART control, YOLOv8, Intel RealSense RGB-D sensing, hand-eye calibration and IKPy inverse kinematics for object grasping.
Selected Academic Projects
Nature Kit | Industrial Design Engineering · January 2025 - June 2025
Designed an interactive early-childhood product through museum/nursery research; developed manufacturing-ready CAD, a front-end UI, generative storytelling and business/cost analysis.
Statistical Machine Learning | Data Science · April 2024 - June 2024
Achieved 81.7% GDP-classification test accuracy with Random Forest/XGBoost; used GridSearchCV, RFE and UMAP to tune models, select features and inspect separability.
Teaching And Leadership
Undergraduate Teaching Assistant | Imperial College London · April 2026 - June 2026
Ran Electronics 1 labs, supported analogue/digital circuit debugging and marked lab work with individual feedback.
Motorsports Society Race Manager (2023-2024): led BUKC 2024 24-hour race logistics and strategy. DesSoc Workshop Manager (2024-2025). Melon Band bassist and Robotics Society member (2024-present).
Technical Skills
Robotics and vision: ROS / ROS2, computer vision, robotic manipulation, teleoperation, MuJoCo, Isaac Gym, RoboCasa.
Programming: Python, C++, JavaScript, MATLAB, PyTorch, OpenCV, Open3D.
Design and tools: Blender, Fusion 360, Arduino IDE, Git, Linux, LaTeX. Languages: English (TOEFL 112), Mandarin Chinese (native).
Interests: Bass, keyboard, Formula 1, racing, film photography, model aircraft and cars, PC hardware.