Two new papers on open-world TAMP and learning symbolic world models from demonstration accepted at RA-L.