State and Trajectory Estimation of Tensegrity Robots via Factor Graphs and Chebyshev Polynomials

This project is a specialized research implementation targeting the niche field of tensegrity robotics. While tensegrity robots offer unique advantages in compliance and durability, their estimation is notoriously difficult due to underconstrained dynamics and non-linearities. The use of factor graphs (likely building on frameworks like GTSAM) combined with Chebyshev polynomials for continuous-time trajectory representation is a sophisticated approach for this specific domain. With 0 stars but 7 forks within 5 days of appearance, there is immediate interest within a small research community, likely related to the arXiv paper submission. However, its defensibility is low as it serves as a reference implementation for a specific paper rather than a general-purpose tool. Frontier labs are unlikely to compete here as the hardware paradigm (tensegrity) is currently on the periphery of mainstream robotics compared to humanoids or quadrupedal systems. The primary risk is displacement by newer neural state estimation techniques or differentiable physics engines (like Brax or Isaac Gym) that could solve these problems more holistically in simulation.