DoReMi: Grounding Language Model by Detecting and Recovering from Plan-Execution Misalignment | IEEE Conference Publication | IEEE Xplore