Ideas without budget limits? Well, I would start with those three:

First, create a home lab with all the required equipment to reverse engineer what's inside two wireless remotes I have here. So basically, sniffing equipment for both the RF signal and what's inside the "black boxes" on the PCB. I would imagine stuff like microscope, radioscope, current sensors that can sense at the "nanometer" scale, even tool that could read embedded flash memory. Yes, that sounds a lot like the lab that the IRCGN has with its 3D X-Ray tomography and laser abrasion bench: 
https://next.ink/6513/108071-la-nouvelle-arme-anti-cryptographie-gendarmerie/
https://www.pressreader.com/france/gq-france/20170517/285860142970333

That might sound a bit too much craziness, so maybe the second idea would be more feasible. I would like to add a LED matrix on my roof tiles so that I can display images/animations whenever I feel like it. Initially, I though of using a 8x8 array of WS2812B LEDs with light guides that would get out just at the corner of a few tiles. But with 20cm between pixels, that would not be really easy to look at. So, with unlimited budget, I would actually create roof tiles that embed a LED array itself, with like 5 centimeters between LEDs so that the overlap area between tiles does not "break" the array. Pretty much like the LED curtains that are now ubiquitous on TV sets or concert shows, but embedded in the tiles with resistance to temperatures from -40°C to +150°C (yes, a tile gets hot).

And finally, one that would be more practical is the creation of a robot that could autonomously pick cherries. Yes, I know, picking fruit robots already exist, from the "shaking tree" for olives to "hand snatch" for apples, but cherries are uniquely challenging. Indeed, the fruit must not fall more than 5 centimeters, be picked at the right time (sometimes 12 hours delay is too much) and with the tail because a cherry without it rots much faster. The computer vision is the easiest part, with the neat detail that it must happen at up to 10 meters in the air. Reaching that far by human task force is already a challenge, but a robot that does all this without "dropping" the fruit would be quite a challenge and require lots of experiments. Like, should I use a "fabric chute", a conveyor belt, a set of mini chair lifts? and should it be a retractable arm as in TV cranes, an arm with multiple folding segments, or even a snake like arm that has lots of flexibility?
And after all that has been figured out, how much would it cost to make it a "viable product"? Probably too much anyway.

Ah well, I guess I'll go back to smaller endeavors.

Ideas without budget limits? Well, I would start with those three:

First, create a home lab with all the required equipment to reverse engineer what's inside two wireless remotes I have here. So basically, sniffing equipment for both the RF signal and what's inside the "black boxes" on the PCB. I would imagine stuff like microscope, radioscope, current sensors that can sense at the "nanometer" scale, even tool that could read embedded flash memory. Yes, that sounds a lot like the lab that the IRCGN has with its 3D X-Ray tomography and laser abrasion bench: 
https://next.ink/6513/108071-la-nouvelle-arme-anti-cryptographie-gendarmerie/
https://www.pressreader.com/france/gq-france/20170517/285860142970333

That might sound a bit too much craziness, so maybe the second idea would be more feasible. I would like to add a LED matrix on my roof tiles so that I can display images/animations whenever I feel like it. Initially, I though of using a 8x8 array of WS2812B LEDs with light guides that would get out just at the corner of a few tiles. But with 20cm between pixels, that would not be really easy to look at. So, with unlimited budget, I would actually create roof tiles that embed a LED array itself, with like 5 centimeters between LEDs so that the overlap area between tiles does not "break" the array. Pretty much like the LED curtains that are now ubiquitous on TV sets or concert shows, but embedded in the tiles with resistance to temperatures from -40°C to +150°C (yes, a tile gets hot).

And finally, one that would be more practical is the creation of a robot that could autonomously pick cherries. Yes, I know, picking fruit robots already exist, from the "shaking tree" for olives to "hand snatch" for apples, but cherries are uniquely challenging. Indeed, the fruit must not fall more than 5 centimeters, be picked at the right time (sometimes 12 hours delay is too much) and with the tail because a cherry without it rots much faster. The computer vision is the easiest part, with the neat detail that it must happen at up to 10 meters in the air. Reaching that far by human task force is already a challenge, but a robot that does all this without "dropping" the fruit would be quite a challenge and require lots of experiments. Like, should I use a "fabric chute", a conveyor belt, a set of mini chair lifts? and should it be a retractable arm as in TV cranes, an arm with multiple folding segments, or even a snake like arm that has lots of flexibility?
And after all that has been figured out, how much would it cost to make it a "viable product"? Probably too much anyway.

Ah well, I guess I'll go back to smaller endeavors.