Uploaded October 2025 | Updated September 2026, 3 days ago
In this talk, Yichen presents his research experience developing robotic systems that work both in simulation and on real hardware. He discusses three key projects that demonstrate his technical experience: (1) developing VR teleoperation systems for controlling two Boston Dynamics Spot robots, where he took charge of the ROS stack, aligned the point clouds from two robots, and improved network throughput; (2) creating composable manipulation skills for dual-arm robots, developing jar opening and spreading skills that transfers from simulation to real hardware; and (3) building a task and motion planner in Isaac Sim using off-the-shelf task planners and motion planners, and creating a data curation pipeline to collect trajectories from the task and motion plan. Across these projects, he has tackled the practical challenges of building reliable robot systems in simulation and on real hardware. He discusses the technical problems he encountered when working with different robotic platforms (Spot, Franka FR3, Kuka iiwa) and how he has used tools like ROS, PyTorch, and Isaac Lab to build the systems. The talk concludes with what he has learned from this experience and the technical directions he is most excited to pursue next.
Yichen Wei finished his master’s degree in Computer Science at Brown University in May 2025, advised by professors George Konidaris and Stefanie Tellex. He has participated in research on robot teleoperation, reinforcement learning, and Task and Motion Planning. At Brown University, he has set up multiple robot systems, including the Boston Dynamics Spot and a Dual-arm Kuka system. He is interested in creating general hierarchical learning and planning systems to solve long-horizon tasks.
In this talk, Yichen presents his research experience developing robotic systems that work both in simulation and on real hardware. He discusses three key projects that demonstrate his technical experience: (1) developing VR teleoperation systems for controlling two Boston Dynamics Spot robots, where he took charge of the ROS stack, aligned the point clouds from two robots, and improved network throughput; (2) creating composable manipulation skills for dual-arm robots, developing jar opening and spreading skills that transfers from simulation to real hardware; and (3) building a task and motion planner in Isaac Sim using off-the-shelf task planners and motion planners, and creating a data curation pipeline to collect trajectories from the task and motion plan. Across these projects, he has tackled the practical challenges of building reliable robot systems in simulation and on real hardware. He discusses the technical problems he encountered when working with different robotic platforms (Spot, Franka FR3, Kuka iiwa) and how he has used tools like ROS, PyTorch, and Isaac Lab to build the systems. The talk concludes with what he has learned from this experience and the technical directions he is most excited to pursue next.
Yichen Wei finished his master’s degree in Computer Science at Brown University in May 2025, advised by professors George Konidaris and Stefanie Tellex. He has participated in research on robot teleoperation, reinforcement learning, and Task and Motion Planning. At Brown University, he has set up multiple robot systems, including the Boston Dynamics Spot and a Dual-arm Kuka system. He is interested in creating general hierarchical learning and planning systems to solve long-horizon tasks.
