Xuewei Zhang (张学伟)

2024.01 - Present   |   Research on Shared Environment Modeling and Cooperative Behavior Planning

With the objective of enabling POU (Point-of-Use) material picking and distribution on production and assembly lines via human-robot arm collaboration.

Personal Contributions:

• Developed a high-frequency (>100Hz) motion planning algorithm for a 7-DOF robotic arm using OMPL, enabling robust collaborative behavior in dynamic environments.
• Designed a POU material picking and distribution process using a custom API, allowing real-time coordination with human operators and the material inventory system.

View Demo on Youtube

2024.01 - 2024.04   |   Terrain-Adaptive Low-Altitude Path Planning for a Fixed-Wing UAV

Aimed at enabling covert low-altitude flight of a fixed-wing UAV in mountainous terrain. The project involved generating safe waypoints based on Digital Elevation Maps (DEM) while under dynamic and terrain-following constraints, including a cruise speed of 0.6 - 0.9 Mach, waypoint intervals over 2.5 km, and terrain clearance under 100 m.

Personal Contributions:

• Modeled terrain-following constraints and flight performance limits to ensure physically feasible and safe low-altitude trajectories over mountainous terrain.
• Developed a terrain-following route planner for a fixed-wing UAV using the Time Elastic Band (TEB) algorithm and g2o-based trajectory optimization within the defined safe flight envelope.
• Deployed the planner on a ROS platform with UE4-based simulation for real-time visualization and performance validation.

View Demo on Youtube

2023.10 - 2024.04   |   Hardware-in-the-loop Simulation System for UAV Swarm Tasks in Airport and Port Operations

This project aimed to develop a high-fidelity hardware-in-the-loop (HIL) simulation system for UAV swarms performing sequential mission tasks in airport and port environments. The system enables integrated verification of core swarm capabilities, including communication networking, formation keeping, task allocation, collision avoidance, trajectory planning, and performance evaluation. A Unity-based 3D visualization interface was also developed for real-time monitoring.

Personal Contributions:

• Designed a coalition game-based task allocation algorithm for UAV swarms, enabling rapid and autonomous decision-making in dynamic, time-critical scenarios.
• Developed a scalable collision avoidance and trajectory planning framework supporting large-scale swarm simulations with up to 256 UAVs.
• Implemented a finite state machine (FSM) integrated with a behavior tree framework for sequential, modular coordination of swarm task modules.
• Built a HIL simulation platform using 32 NVIDIA Jetson Orin NX running PX4-based flight units for real-time control of physical nodes, along with 224 virtual UAVs simulated on a PC.

2023.08   |   The 3rd Air Force "Unmanned Contenders" Intelligent UAV Swarm System Challenge

The challenge aimed to evaluate intelligent UAV swarm coordination and precision strike capabilities in dynamic mission scenarios. A group of 3 - 5 quadrotors autonomously searched for designated targets within a competition zone. Upon detecting targets, strike commands were relayed to fixed-wing UAVs loitering overhead, which then executed simulated attacks.

Personal Contributions:

• Designed and built quadrotor platforms from the ground up, integrating PX4 firmware and tuning PID parameters for optimal flight performance. Each quadrotor is equipped with a C-RTK 9Ps GNSS module, Intel RealSense D435i depth camera, CUAV V5+ flight controller, and an NVIDIA Jetson Xavier onboard computer for autonomous operation.
• Developed a custom ground control station using ROS-Rviz and Qt, supporting real-time UAV state monitoring, mission visualization, and unified command dispatch for multi-UAV coordination.
• Developed a Coverage Path Planning Algorithm (CPPA) based on the WGS-84 for accurate waypoint generation, and transmitted waypoints to PX4 via MAVROS using the MAVLink protocol for onboard execution.

View Demo on Youtube

2022.10 - 2023.10   |   Decentralized Multi-UAV Cooperative Exploration in GPS-Denied Environments

With the goal of enabling decentralized multi-UAV cooperative exploration in large-scale forest environments (~5000 m³) under GPS-denied conditions.

Personal Contributions:

• Designed and built a SLAM-capable UAV platform equipped with a Livox Mid-360 LiDAR, Pixhawk flight controller, network communication module, and an onboard Nvidia Jetson Orin NX for autonomous localization and navigation.
• Implemented a high-speed concurrent communication bridge based on the ZeroMQ framework to support TCP/IP and UDP message relaying among ROS1 nodes, serving as a lightweight and efficient alternative to the ROS1 MultiMaster communication setup.

View Demo on Youtube

2022.07   |   “Feihang Cup” UAV Swarm Challenge, 1st Intelligent Equipment & Systems Design Competition

The competition was conducted in a UE4-based simulation environment. Teams interfaced with the virtual scene via ROS nodes to receive real-time environmental data and programmed four UAVs to perform autonomous navigation and collaborative target search.

Personal Contributions:

• Developed an incremental topological map to represent explored and unexplored regions, enabling lightweight communication for consistent environmental maintenance.
• Designed a hierarchical motion planning algorithm using the Model Predictive Path Integral (MPPI) framework for UAVs to ensure safe and efficient navigation in obstacle-dense environments.

View Demo on Youtube

2020.05 - 2022.04   |   Mission Planning for Coaxial Twin-Rotor UAV Swarm Systems

A coaxial twin-rotor UAV swarm (≥12 units) forms a coordinated formation to perform situational awareness, target reconnaissance, and dynamic strike missions.

Personal Contributions:

• Designed and implemented formation keeping and formation reconfiguration algorithms for UAV swarms, achieving reliable and smooth transitions between formations with precise relative positioning and collision avoidance capabilities.
• Implemented coordinate transformation between WGS-84 and Cartesian coordinate systems, and established a unified Home point setting for the UAV swarm system.

View Demo on Youtube