# Jianyu's Letter

I am writing to enthusiastically recommend Yen-Jen Wang, a distinguished master’s student who has been under my supervision since September 2021. I can confidently and proudly say that among all the students I have recommended this year, Yen-Jen stands out as the most exceptional.

Yen-Jen's ability to learn and implement new skills is extraordinary. Despite having no background in robotics prior to joining Tsinghua University, he swiftly mastered the Robotics Operating System (ROS) and adeptly completed a project involving a mobile manipulator in just a few days. His commitment to mastering new fields of study is truly commendable.

In early 2022, our lab embarked on developing a humanoid robot named Xiaoxing, a project encompassing mechanical design, embedded system development, and control algorithm creation. Yen-Jen was involved in every stage of this development, with a particular focus on sim-to-real transfer using reinforcement learning (RL). This task, especially challenging for a new robot like Xiaoxing, required not only technical know-how but also immense perseverance and patience. Impressively, Yen-Jen excelled in this role, enabling our humanoid robot to walk autonomously through RL. He also significantly deepened his knowledge in robotics, covering both hardware and software aspects, and enhanced his communication skills through effective collaboration with multidisciplinary experts.

Furthermore, Yen-Jen's research approach is marked by keen critical thinking and a collaborative nature. He meticulously analyzes situations and boldly explores innovative ideas. In April 2023, he identified the crucial need for aligning high-level planning with low-level actions in long-term robotic tasks. Building on this insight, he developed the concept of "DoReMi: Grounding Language Model by Detecting and Recovering from Plan-Execution Misalignment," showcasing his inventive approach to problem-solving. During his summer 2023 visiting research at Professor Koushil Sreenath's Hybrid Robotics Lab at UC Berkeley, Yen-Jen questioned the existing applications of Language Model Learning (LLM) in robotics and pioneered a project utilizing LLM for robotic low-level control. His ability to collaborate is also highly valued, as evidenced by his peers' eagerness to work with him on multiple projects.

Additionally, Yen-Jen is extremely thorough and dependable in his work. I have complete confidence in his abilities whenever I assign him tasks. As a teaching assistant for the "Advanced Topics for Robotics" course in Spring 2023, his commitment to student success was apparent, often offering support late into the night. His meticulous planning and hospitality during the World Artificial Intelligence Conference 2023 in Shanghai, especially in hosting Professors Anca Dragan and Koushil Sreenath, received widespread acclaim.

In summary, Yen-Jen's skills, dedication, and teamwork make him an outstanding candidate for any Ph.D. program. I wholeheartedly recommend him without any reservations and eagerly anticipate his future contributions to the field of robotics.

<!-- Polish up:

I am writing to wholeheartedly recommend Yen-Jen Wang, a distinguished master’s student who has been under my supervision since September 2021. I can confidently and proudly say that among all the students I have recommended this year, Yen-Jen stands out as the most exceptional.

Yen-Jen possesses remarkable learning and implementation skills. Despite having no prior experience in robotics before joining Tsinghua University, he quickly mastered the Robotics Operating System (ROS) and skillfully completed a project involving a mobile manipulator—all within a few days. His dedication to adapting to new areas of study is truly remarkable. In early 2022, we began building our own humanoid robot named Xiaoxing in our lab. This big project involved mechanical design, embedded development, and control algorithm design. Yen-Jen participated in the entire development process and mainly focused on sim-to-real transfer of reinforcement learning (RL). This task, particularly challenging for a new robot, like Xiaoxing, demands not only technical expertise but also remarkable perseverance and patience. What is amazing is that Yen-Jen worked hard during this process and recently completed the task, enabling our humanoid robot to walk through RL. Simultaneously, he deepened his knowledge in robotics from the aspect of both hardware and software and honed his communication skills, effectively collaborating with experts across various disciplines.

In addition, Yen-Jen's research approach is characterized by his ability to think critically and work well with others. He analyzes things in detail and isn't afraid to try out new ideas. In April 2023, he discovered the alignment between high-level plan and low-level action with current long-horzon robotic tasks is vital. Building on this insight, he proposed the basic idea of "DoReMi: Grounding Language Model by Detecting and Recovering from Plan-Execution Misalignment" to enhance this problem, demonstrating his innovative thinking in the field. In the summer of 2023, Yen-Jen went to Professor Koushil Sreenath's Hybrid Robotics lab at UC Berkeley for visiting research on my recommendation. During that time, he doubted the previous usage of LLM in the robotics area and worked on a project that leverage LLM for robotic low-level control. It's worth to mention that Yen-Jen is also a sought-after collaborator, as evidenced by his peers' desire to continue working with him after each project.

Last but not least, I must mention that Yen-Jen is very careful and reliable in his work. As long as I asign everything to him, I don't need to worry. As a teaching assistant for the "Advanced Topics for Robotics" course in Spring 2023, Yen-Jen was exceptional. His dedication to student success was evident through his willingness to support them well into the night. Additionally, his meticulous planning and hospitality during the World Artificial Intelligence Conference 2023 in Shanghai, as he hosted Professors Anca Dragan and Koushil Sreenath, was highly praised.

To conclude, Yen-Jen's capabilities, dedication, and collaborative spirit make him an exemplary candidate for any Ph.D. program. I confidently recommend him without reservation, anticipating his significant future contributions to the field of robotics. -->