Engineering autonomous systems from the ground up — where perception, learning, control, and real hardware meet.
From embedded electronics and mechanical integration to vision, planning, and machine learning, our work explores how complete robotic systems can operate reliably beyond a simulation.
Applied projects across manipulation, mobility, and autonomy.
PERCEPTION-TO-GRASPAn industrial automation concept built around a single overhead RGB-D sensor, a 6-DoF manipulator, and a dexterous hand. The planned pipeline investigates grasp candidate generation, motion planning, and residual reinforcement learning for near-contact refinement.
Discuss this work
MOBILE AUTONOMYA mobile robotics project spanning hardware integration, suspension design, power electronics, navigation, and elevator-aware operation. The system brings together embedded components and ROS 2-based autonomy.
Discuss this work
LEARNING-BASED CONTROLResearch prototypes evaluated in CARLA, using camera inputs for steering and longitudinal control. The study explores behavior cloning, multi-scale visual representations, and reinforcement-learning-based policy refinement.
Discuss this workMethods and technologies used across our projects.
Computer vision, RGB-D sensing, object understanding, and visual representations.
Robot motion planning, trajectory tracking, kinematics, and closed-loop control.
Imitation learning, reinforcement learning, and simulation-to-reality workflows.
Embedded systems, motor control, electronics, and robotic hardware integration.
Our development direction is to connect learning-based intelligence with dependable physical robot systems — especially where perception, manipulation, and control must work together under real-world constraints.
Research and prototypes shown on this site are at different stages of development. Technical details and demonstrations can be shared on request.
Interested in robotics research, engineering collaborations, or learning more about the projects? Get in touch.