Mixed-Integer Programming for Signal Temporal Logic With Fewer Binary Variables

Signal Temporal Logic (STL) provides a convenient way of encoding complex control objectives for robotic and cyber-physical systems. The state-of-the-art in trajectory synthesis STL is based on Mixed-Integer Convex Programming (MICP). MICP approach sound complete, but has limited scalability due to exponential complexity the number binary variables. In this letter, we propose more efficient STL. Our new insight that disjunction can be encoded using logarithmic variables conjunction without We demonstrate simulation examples our proposed significantly outperforms long specifications. Open-source software available: https://stlpy.readthedocs.io .