Shiqi Liu

Papers

Total Citations

H-Index

About

Shiqi Liu is at the forefront of advancing robotic autonomy, with a primary focus on bridging the critical gap between simulation and real-world deployment. A key contribution is the development of a differentiable causal discovery framework that systematically identifies and corrects the "sim-to-real" gap in robot dynamics, a fundamental challenge that has long hindered the transfer of learned policies from virtual environments to physical hardware. This work, garnering early citations for its novel approach, provides a principled method for understanding *why* a simulation fails, rather than merely treating the symptoms. Liu also tackles the grand challenge of multi-agent loco-manipulation, pioneering learning-based methods for long-horizon quadrupedal pushing—a task demanding both robust locomotion and coordinated manipulation of large objects. This research directly addresses limitations in real-world applications like search and rescue and industrial automation. Furthermore, Liu has explored the creative use of large language models to imbue robots with tool-use capabilities for tasks requiring implicit physical reasoning and long-term planning. With a rapidly growing citation record, Shiqi Liu is establishing a reputation for tackling the hardest problems in embodied AI, from causal reasoning to complex physical interaction.