Sim2Sim Evaluation of a Novel Data-Efficient Differentiable Physics Engine for Tensegrity Robots

This project is a research artifact associated with a 2020 paper. With 0 stars and 3 forks over nearly 2,000 days, it has failed to gain any meaningful community traction or adoption. While differentiable physics for soft/tensegrity robotics was a novel research area around its release, the space has since been dominated by more robust, general-purpose differentiable simulators like NVIDIA Warp, Google Brax, and JAX-based physics engines. The defensibility is extremely low as it is a specialized reference implementation for a niche robot morphology (tensegrity) that has not been maintained. Frontier labs are unlikely to compete directly in tensegrity physics, but the general-purpose simulation platforms they (and others like NVIDIA) provide make this specific engine obsolete. The 'sim2sim' methodology is now standard practice in robotics research, offering no unique moat.