Research direction

2D Vision

At MVRLab, our 2D Vision research focuses on enabling robots to perceive and interpret the world through standard visual sensors. By combining advanced image processing with deep learning, we develop algorithms that allow robots to recognize objects, detect features, and understand scenes in real time.

We explore key topics such as:

• Object Detection and Recognition – enabling robots to identify tools, components, and environments with high accuracy.

• Pose Estimation and Tracking – allowing robots to follow human movements or track dynamic objects safely.

• Scene Understanding – teaching robots to interpret complex environments for navigation and manipulation.

• Low-Cost Vision Systems – developing lightweight, affordable solutions for industrial and service applications.

Our work in 2D Vision is not only about perception, but also about integration with planning and control. By giving robots the ability to “see” in two dimensions, we empower them to interact intelligently with their surroundings, collaborate with humans, and adapt to diverse real-world scenarios.

Students and researchers joining this direction will gain hands-on experience with camera systems, neural networks, and robotic platforms. They will contribute to projects that bridge AI and robotics—developing innovative solutions that advance both academic research and industrial applications.

3D Vision

Our 3D Vision research goes beyond flat images, enabling robots to perceive depth, shape, and geometry. By leveraging stereo cameras, depth sensors, and 3D point cloud processing, we equip robots with the ability to interact in complex environments with higher accuracy and reliability.

Key research areas include:

• 3D Object Recognition – identifying and localizing objects in cluttered or dynamic scenes.

• Point Cloud Processing – using advanced filtering, segmentation, and feature extraction for accurate perception.

• 3D Mapping and SLAM – creating robust environment maps for navigation and manipulation.

• Human–Robot Interaction in 3D – enabling robots to understand gestures, movements, and collaborative tasks.

By integrating 3D perception with motion planning and control, our robots can navigate tight spaces, avoid obstacles, and perform precision tasks. Students and researchers in this direction gain practical experience with state-of-the-art 3D cameras, LIDAR systems, and robotics platforms, contributing to solutions that advance smart manufacturing, service robotics, and beyond.

AI for Robotics

At MVRLab, Artificial Intelligence (AI) for Robotics is about making robots more intelligent, adaptive, and capable of learning from experience. We focus on applying machine learning, deep learning, and reinforcement learning to enhance robot autonomy across perception, decision-making, and control.

Our research spans:

• Learning-Based Control – using AI to optimize control strategies in dynamic environments.

• Reinforcement Learning for Skills Acquisition – enabling robots to learn complex behaviors through trial and error.

• Human-Robot Collaboration – developing AI methods that allow robots to predict human intent and work safely alongside people.

• Sim2Real Transfer – training robots in simulation and transferring knowledge effectively to the real world.

With AI, our robots move beyond rigid programming, becoming adaptive agents capable of handling uncertainty, learning new tasks, and improving over time. Students and researchers in this area gain expertise in combining theory with practice, working on projects that connect AI algorithms with real robotic platforms to push the frontier of intelligent automation.