One thing to try: Pick one idea. Book recurring time (does not have to be that long time) to work on the project. During that booked time, do something that pus

One thing to try: Pick one idea. Book recurring time (does not have to be that long time) to work on the project. During that booked time, do something that pushes the project forward. Draw concrete UI diagrams, write real code, design a few database tables. Do not spend that time on "research". Start by keeping those timeslots so short you must get to work quickly to finish at least one thing (artificial deadline looming there). Finish each slot by writing a note about where to continue next. Or leave code broken or a unit test broken. Next time you come back, you should have something simple and obvious to start.